Vere scire est per causas scire
Human of the Future & Artificial Intelligence
In
order for the political programs of the parties to be in line with the spirit
of the times, in order for the ideological attitudes to acquire a progressive nature,
it is essential to take into account the trends of the development of modern
civilization!
One of the dominant attributes of the
progress of modern society is the increasingly closer confluence of the human
and digital technology, a symbiosis of biological and artificial intelligence. This
results in a radical change of the entire civilization, while the human,
step by step, approaches the greatness of God.
This is a daunting challenge to the
morality of today’s human and at the same time a real existential threat! The
entire future of civilization depends and will depend on whether we will be
able to use the opportunities offered by modern science in conformity with the
criteria of humanism.
The publications presented here allow us to
predict the prospects for the development of society and open slightly the
curtain to the future in the presence of artificial intelligence. The future
prospects of each person and the whole society depend on how and in what way we
will be able to use the opportunities created by artificial intelligence!
Artificial intelligence
can be not only a valuable assistant, but also a dangerous enemy.
In the wrong hands,
artificial intelligence can become a means of manipulation.
by Kai-Fu Lee ,
Chen Qiufan
In a groundbreaking blend of science and
imagination, the former president of Google China and a leading writer of
speculative fiction join forces to answer an urgent question: How will
artificial intelligence change our world over the next twenty years?
AI will be the defining issue of the twenty-first century, but many people know
little about it apart from visions of dystopian robots or flying cars. Though
the term has been around for half a century, it is only now, Kai-Fu Lee argues,
that AI is poised to upend our society, just as the arrival of technologies
like electricity and smart phones did before it. In the past five years, AI has
shown it can learn games like chess in mere hours--and beat humans every time.
AI has surpassed humans in speech and object recognition, even outperforming
radiologists in diagnosing lung cancer. AI is at a tipping point. What comes
next?
Within two decades, aspects of daily life may be unrecognizable. Humankind
needs to wake up to AI, both its pathways and perils. In this provocative work
that juxtaposes speculative storytelling and science, Lee, one of the world's
leading AI experts, has teamed up with celebrated novelist Chen Qiufan to
reveal how AI will trickle down into every aspect of our world by 2041. In ten
gripping narratives that crisscross the globe, coupled with incisive analysis,
Lee and Chen explore AI's challenges and its potential:
- Ubiquitous AI that knows you better than you know yourself
- Genetic fortune-telling that predicts risk of disease or even IQ
- AI sensors that creates a fully contactless society in a future
pandemic
- Immersive personalized entertainment to challenge our notion of
celebrity
- Quantum computing and other leaps that both eliminate and
increase risk
By gazing toward a not-so-distant horizon, AI 2041 offers
powerful insights and compelling storytelling for everyone interested in our
collective future.
https://www.goodreads.com/en/book/show/56377201-ai-2041
The Bletchley Declaration
by Countries Attending the AI Safety
Summit, 1-2 November 2023
Artificial Intelligence (AI) presents enormous global opportunities: it has the potential to transform and enhance human wellbeing, peace and prosperity. To realise this, we affirm that, for the good of all, AI should be designed, developed, deployed, and used, in a manner that is safe, in such a way as to be human-centric, trustworthy and responsible. We welcome the international community’s efforts so far to cooperate on AI to promote inclusive economic growth, sustainable development and innovation, to protect human rights and fundamental freedoms, and to foster public trust and confidence in AI systems to fully realise their potential.
AI systems are already
deployed across many domains of daily life including housing, employment,
transport, education, health, accessibility, and justice, and their use is
likely to increase. We recognise that this is therefore a unique moment to act
and affirm the need for the safe development of AI and for the
transformative opportunities of AI to be used for good and for all,
in an inclusive manner in our countries and globally. This includes for public
services such as health and education, food security, in science, clean energy,
biodiversity, and climate, to realise the enjoyment of human rights, and to
strengthen efforts towards the achievement of the United Nations Sustainable
Development Goals.
Alongside these
opportunities, AI also poses significant risks, including in those
domains of daily life. To that end, we welcome relevant international efforts
to examine and address the potential impact of AI systems in existing
fora and other relevant initiatives, and the recognition that the protection of
human rights, transparency and explainability, fairness, accountability,
regulation, safety, appropriate human oversight, ethics, bias mitigation,
privacy and data protection needs to be addressed. We also note the potential
for unforeseen risks stemming from the capability to manipulate content or
generate deceptive content. All of these issues are critically important and we
affirm the necessity and urgency of addressing them.
Particular safety risks arise
at the ‘frontier’ of AI, understood as being those highly capable
general-purpose AI models, including foundation models, that could
perform a wide variety of tasks - as well as relevant specific narrow AI that
could exhibit capabilities that cause harm - which match or exceed the
capabilities present in today’s most advanced models. Substantial risks may
arise from potential intentional misuse or unintended issues of control
relating to alignment with human intent. These issues are in part because those
capabilities are not fully understood and are therefore hard to predict. We are
especially concerned by such risks in domains such as cybersecurity and
biotechnology, as well as where frontier AI systems may amplify risks
such as disinformation. There is potential for serious, even catastrophic,
harm, either deliberate or unintentional, stemming from the most significant
capabilities of these AI models. Given the rapid and uncertain rate
of change of AI, and in the context of the acceleration of investment in
technology, we affirm that deepening our understanding of these potential risks
and of actions to address them is especially urgent.
Many risks arising
from AI are inherently international in nature, and so are best
addressed through international cooperation. We resolve to work together in an
inclusive manner to ensure human-centric, trustworthy and responsible AI that
is safe, and supports the good of all through existing international fora and
other relevant initiatives, to promote cooperation to address the broad range
of risks posed by AI. In doing so, we recognise that countries
should consider the importance of a pro-innovation and proportionate
governance and regulatory approach that maximises the benefits and
takes into account the risks associated with AI. This could include
making, where appropriate, classifications and categorisations of risk based on
national circumstances and applicable legal frameworks. We also note the
relevance of cooperation, where appropriate, on approaches such as common
principles and codes of conduct. With regard to the specific risks
most likely found in relation to frontier AI, we resolve to intensify and
sustain our cooperation, and broaden it with further countries, to identify,
understand and as appropriate act, through existing international fora and
other relevant initiatives, including future international AI Safety
Summits.
All actors have a role to
play in ensuring the safety of AI: nations, international fora and other
initiatives, companies, civil society and academia will need to work together.
Noting the importance of inclusive AI and bridging the digital
divide, we reaffirm that international collaboration should endeavour to engage
and involve a broad range of partners as appropriate, and welcome
development-orientated approaches and policies that could help developing
countries strengthen AI capacity building and leverage the enabling
role of AI to support sustainable growth and address the development
gap.
We affirm that, whilst safety
must be considered across the AI lifecycle, actors developing
frontier AI capabilities, in particular those AI systems
which are unusually powerful and potentially harmful, have a particularly strong
responsibility for ensuring the safety of these AI systems, including
through systems for safety testing, through evaluations, and by other
appropriate measures. We encourage all relevant actors to provide
context-appropriate transparency and accountability on their plans to measure,
monitor and mitigate potentially harmful capabilities and the associated
effects that may emerge, in particular to prevent misuse and issues of control,
and the amplification of other risks.
In the context of our
cooperation, and to inform action at the national and international levels, our
agenda for addressing frontier AI risk will focus on:
- identifying AI safety risks of shared
concern, building a shared scientific and evidence-based understanding of
these risks, and sustaining that understanding as capabilities continue to
increase, in the context of a wider global approach to understanding the
impact of AI in our societies.
- building respective risk-based policies across
our countries to ensure safety in light of such risks, collaborating as
appropriate while recognising our approaches may differ based on national
circumstances and applicable legal frameworks. This includes, alongside
increased transparency by private actors developing
frontier AI capabilities, appropriate evaluation metrics, tools
for safety testing, and developing relevant public sector capability and
scientific research.
In furtherance of this
agenda, we resolve to support an internationally inclusive network of
scientific research on frontier AI safety that encompasses and
complements existing and new multilateral, plurilateral and bilateral
collaboration, including through existing international fora and other relevant
initiatives, to facilitate the provision of the best science available for
policy making and the public good.
In recognition of the
transformative positive potential of AI, and as part of ensuring wider
international cooperation on AI, we resolve to sustain an inclusive global
dialogue that engages existing international fora and other relevant initiatives
and contributes in an open manner to broader international discussions, and to
continue research on frontier AI safety to ensure that the benefits
of the technology can be harnessed responsibly for good and for all. We look
forward to meeting again in 2024.
by Jacob Ward
This eye-opening
narrative journey into the rapidly changing world of artificial intelligence
reveals the alarming ways AI is exploiting the unconscious habits of our brains
– and the real threat it poses to humanity.: https://www.goodreads.com/en/book/show/59429424-the-loop
‘Trustworthy AI: A Business Guide for Navigating
Trust and Ethics in AI’
by Bina Ammanathi
The founders of Humans
for AI provide a straightforward and structured way to think about belief and
ethics in AI, and provide practical guidelines for organizations developing or
using artificial intelligence solutions.: https://soundcloud.com/reesecrane/pdfreadonline-trustworthy-ai-a-business-guide-for-navigating-trust-and
‘The New
Fire: War, Peace and Democracy in the Age of AI’
by Ben Buchanan and
Andrew Embry
Combining a sharp
grasp of technology with clever geopolitical analysis, two AI policy experts
explain how artificial intelligence can work for democracy. With the right
approach, technology need not favor tyranny. : https://www.goodreads.com/en/book/show/58329461-the-new-fire
Adversarial
vulnerabilities of human decision-making
November 17, 2020
Significance
“What I cannot efficiently break, I cannot
understand.” Understanding the vulnerabilities of human choice processes allows
us to detect and potentially avoid adversarial attacks. We develop a general
framework for creating adversaries for human decision-making. The framework is
based on recent developments in deep reinforcement learning models and
recurrent neural networks and can in principle be applied to any
decision-making task and adversarial objective. We show the performance of the
framework in three tasks involving choice, response inhibition, and social
decision-making. In all of the cases the framework was successful in its
adversarial attack. Furthermore, we show various ways to interpret the models
to provide insights into the exploitability of human choice.
Abstract
Adversarial examples are carefully crafted
input patterns that are surprisingly poorly classified by artificial and/or
natural neural networks. Here we examine adversarial vulnerabilities in the
processes responsible for learning and choice in humans. Building upon recent
recurrent neural network models of choice processes, we propose a general
framework for generating adversarial opponents that can shape the choices of
individuals in particular decision-making tasks toward the behavioral patterns
desired by the adversary. We show the efficacy of the framework through three
experiments involving action selection, response inhibition, and social
decision-making. We further investigate the strategy used by the adversary in
order to gain insights into the vulnerabilities of human choice. The framework
may find applications across behavioral sciences in helping detect and avoid
flawed choice. https://www.pnas.org/content/117/46/29221
Human Compatible:
Artificial
Intelligence and the Problem of Control
In the popular imagination,
superhuman artificial intelligence is an approaching tidal wave that threatens
not just jobs and human relationships, but civilization itself. Conflict
between humans and machines is seen as inevitable and its outcome all too
predictable.
In this groundbreaking book, distinguished AI researcher Stuart Russell argues
that this scenario can be avoided, but only if we rethink AI from the ground
up. Russell begins by exploring the idea of intelligence in humans and in
machines. He describes the near-term benefits we can expect, from intelligent
personal assistants to vastly accelerated scientific research, and outlines the
AI breakthroughs that still have to happen before we reach superhuman AI. He
also spells out the ways humans are already finding to misuse AI, from lethal
autonomous weapons to viral sabotage.
If the predicted breakthroughs occur and superhuman AI emerges, we will have
created entities far more powerful than ourselves. How can we ensure they
never, ever, have power over us? Russell suggests that we can rebuild AI on a
new foundation, according to which machines are designed to be inherently
uncertain about the human preferences they are required to satisfy. Such
machines would be humble, altruistic, and committed to pursue our objectives,
not theirs. This new foundation would allow us to create machines that are
provably deferential and provably beneficial.
In a 2014 editorial co-authored with Stephen Hawking, Russell wrote,
"Success in creating AI would be the biggest event in human history.
Unfortunately, it might also be the last." Solving the problem of control
over AI is not just possible; it is the key that unlocks a future of unlimited
promise. ..: https://www.goodreads.com/en/book/show/44767248-human-compatible
Rise of AI 2020
Welcome to State
of AI Report 2021
Published by Nathan Benaich and Ian Hogarth on 12
October 2021.
This year’s report
looks particularly at the emergence of transformer technology, a technique to
focus machine learning algorithms on important relationships between data
points to extract meaning more comprehensively for better predictions, which
ultimately helped unlock many of the critical breakthroughs we highlight
throughout…:
https://www.stateof.ai/2021-report-launch.html
Statement on AI Risk
AI
experts and public figures express their concern about AI risk
AI experts, journalists, policymakers, and the public are increasingly
discussing a broad spectrum of important and urgent risks from AI. Even so, it
can be difficult to voice concerns about some of advanced AI’s most severe
risks. The succinct statement below aims to overcome this obstacle and open up
discussion. It is also meant to create common knowledge of the growing number
of experts and public figures who also take some of advanced AI’s most severe
risks seriously.
Mitigating the risk of extinction from AI should be a global priority
alongside other societal-scale risks such as pandemics and nuclear war.
https://www.safe.ai/statement-on-ai-risk#signatories
The Great AI Reckoning
Deep learning has built a brave new world—but now
the cracks are showing
The
Turbulent Past and Uncertain Future of Artificial Intelligence
Is there a way out of AI's boom-and-bust cycle?...:
https://spectrum.ieee.org/special-reports/the-great-ai-reckoning/
Stanford University(link is external):
Gathering Strength, Gathering Storms: The One
Hundred Year Study on Artificial Intelligence (AI100) 2021 Study Panel Report
Welcome to the 2021 Report :
https://ai100.stanford.edu/sites/g/files/sbiybj18871/files/media/file/AI100Report_MT_10.pdf
Artificial Intelligence Index Report 2021
https://aiindex.stanford.edu/wp-content/uploads/2021/11/2021-AI-Index-Report_Master.pdf
A
physicist on why AI safety is ‘the most important conversation of our time’
‘Nothing in
the laws of physics that says we can’t build machines much smarter than us’
by Angela Chen@chengela Aug 29, 2017
Should we be worried about the dangerous potential of artificial
intelligence?
Sort of, says Max Tegmark,
a physicist at the Massachusetts Institute of Technology. Tegmark is a
co-founder of the Future of Life
Institute, a Boston-based research organization that studies global
catastrophic risk, with an especial focus on AI. He’s also the author
of Life 3.0, which is out
today. Life 3.0 outlines the current state of AI safety
research and the questions we’ll need to answer as a society if we want the
technology to be used for good.
Tegmark doesn’t believe that doomsaying Terminator scenarios
are inevitable, but he doesn’t think that we’re doing enough thinking about
artificial intelligence either. And he’s not the only one who’s
concerned. Stephen
Hawking has been urging researchers to pay more attention to AI safety
for years now. Microsoft partnered
with Elon Musk to create OpenAI, an organization dedicated to the same
issue. Musk, who also donated
$10 million to FLI in 2015, also recommended Life
3.0 on Twitter.
The Verge chatted with Tegmark about
his book, what we should be doing, and why he thinks the discussion around AI
is the most important one of our time. This interview has been lightly
edited and condensed for clarity.
What does
the title of the book refer to? What is “Life 3.0” and what were Life 1.0 and
2.0?
Well, I think of life broadly as any process that can retain its
complexity and reproduce. Life 1.0 would be like bacteria. Bacteria are small
atoms put together in the form of simple algorithms that control what they do.
For instance, whenever the bacteria notices there is a higher sugar
concentration in front than in back of it, it comes forward, but if it notices
that there’s less sugar in front of it, then it turns around. But bacteria can
never truly learn anything in its lifetime. The only way that bacteria can
gradually get better software, or learn, is through evolution over many
generations.
We humans are what I call “life 2.0.” We still have our hardware, or
bodies, largely designed by evolution, but we can learn. We have enormous power
to upload new “software” into our minds. For example, if you decide you want to
become a lawyer, you can go to law school, and law school involves uploading
new algorithms into your brain so that now suddenly you can have the expertise
of the lawyer. It’s this ability to design our own software rather than having
to wait for evolution to give it to us that enable us to dominate this planet
and create modern civilization and culture. Cultural evolution comes precisely
from the fact that we can copy ideas and knowledge from other people in our
lifetime.
Life 3.0 is life that fully breaks free of its evolutionary shackles and
is able to design not only its software, but its hardware. Put another way, if
we create AI which is at least as smart as us, then it can not only design its
own software to make itself learn new things, but there’s always an attempt to
swap up upgraded memory to remember a million times more stuff, or get more
computing power. In contrast, humans can put in artificial pacemakers or
artificial knees, but we can’t change anything truly dramatic. You can never
make yourself a hundred times taller or a thousand times faster at thinking.
Our intelligence is made of squishy biological neurons and is fundamentally
limited by how much brain mass fits through our mom’s birth canal, but
artificial intelligence isn’t.
Some
people are still skeptical that superintelligence will happen at all, but you
seem to believe strongly that it will, and it’s just a matter of time. You’re a
physicist, what’s your take from that perspective?
I think most people think of intelligence as something mysterious and
limited to biological organisms. But from my perspective as a physicist,
intelligence is just simply information processing performed by elementary
particles moving around according to the laws of physics. Nothing in the laws
of physics that says we can’t build machines much smarter than us, or that
intelligence needs to be built from organic matter. I don’t think there’s any
secret sauce that absolutely requires carbon atoms or blood.
I had a lot of fun in the book thinking about what are the ultimate
limits of the laws of physics on how smart you can be. The short answer is that
it’s sky-high, millions and millions and millions of times above where we are
now. We ain’t seen nothing yet. There’s a huge potential for our universe to
wake up much more, which I think is an inspiring thought, coming from a
cosmology background.
I know
that FLI does work with issues like nuclear disarmament, but it made spreading
the word about AI safety its first major goal. Similarly, you believe that the
conversation around AI safety is “the most important conversation.” Why? Why is
it more important than, say, climate change?
We’ve done a lot for nuclear war risk reduction, but the question of a
good future with AI is absolutely more important than all of these other
things. Take climate change: Yes, it might create huge problems for us in 50
years or 100 years, but many leading AI researchers think that superhuman
intelligence will arrive before then, in a matter of decades.
That’s obviously a way bigger deal than climate change. First of all, if
that happens, it would utterly transform life as we know it. Either it helps
you flourish like never before or becomes the worst thing that ever happened to
us. And second, if it goes well, we could use it to solveclimate
change and all our other problems. If you care about poverty, social justice,
climate change, disease — all of these problems stump us because we’re not
smart enough to figure out how to solve them, but if we can amplify our own
intelligence with machine intelligence, far beyond ours, we have an incredible
potential to do better.
So, it’s something which is different from all the other things on your
list in that there’s not just possible downsides, but huge possible upsides in
that it can help solve all the other problems. Cancer, for example, or disease
more generally — cancer can be cured, it’s just that we humans haven’t been
smart enough to figure out how to deal with it in all cases. We’re limited by
our own intelligence in all the research we do.
There’s a
fairly wide spectrum of attitudes on this topic, from the skeptics to the
utopians. Where do you put yourself?
I’m optimistic that it’s possible to create superhuman intelligence, and
I’m also optimistic in that we can create a great future with AI. But I’m
cautious in the sense that I don’t think it’s guaranteed. There are crucial
questions we have to answer first for things to go well, and they might take 30
years to answer. We should get cracking on them now, not the night before some
dudes decide to switch on their superintelligence.
What
questions? You said you’re not focused in what you call “near-term” questions, like
how automation is going to affect jobs.
There’s so much talk all the time now about job automation, people tend
to forget that it’s important to also look at what comes next. I’m talking
about questions like, how do we transform today’s easily hackable computers
into robust AI systems? How can you make AI systems understand our goals as
they get ever-smarter?
When your computer crashes, it’s annoying, you lost an hour of work, but
it wouldn’t be as funny if that computer were controlling the airplane you were
flying in or the nuclear arsenal of the Untied States.
What goals should AI have? Should it be the goals of some American
computer programmers, or the goals of ISIS or of people in the Middle Ages?
What kind of society can we create? Look how much polarization is there in the
US right now.
If we don’t know what we want, we’re less likely to get it. You can’t
leave this conversation just to tech geeks like myself, either, because the
question of what sort of society we create is going to affect everybody.
A lot of
the initiatives you discuss are big-picture — like how laws need to be updated
to keep up with AI. But what about the average person? What are we supposed to
do?
Talking is a great start. The fact that none of the two presidential
candidates in our last election talked about AI at all, even though I think
it’s the most important issue facing us, is a reflection of the fact that
people aren’t talking about it and therefore don’t care about it when they
vote.
When will
you consider yourself “successful” in making sure we’ve had this conversation?
Look at it this way: we have billions and billions of dollars now
invested in making AI more powerful and almost nothing in AI safety research.
No governments of the world have said that AI safety research should be an
integral part of their computer science funding, and it’s like, why would you
fund building nuclear reactors without funding nuclear reactor safety? Now
we’re funding AI research with no budget in sight for AI safety. I’m certainly
not going to say that we have enough conversation about this until this, at
least, changes. Every AI researcher I know thinks it would be a good idea to
have more funding for this. So I’d say, we’re successful when things like this
are going a little bit in the right direction.
|
“Computing Machinery and Intelligence”
Protecting Against AI’s Existential Threat
How to avoid the nightmare
scenario of artificial intelligence? According to researchers from Elon Musk’s
OpenAI, the trick is teaching machines to keep our interests in mind
Ilya
Sutskever and Dario Amodei
Oct. 18, 2017
On July 8, 2017, an AI
system built by our research company, OpenAI, beat a semipro human player in
solo matches of a battle arena video game called Dota 2. One month later, the
same AI system beat a professional gamer ranked in the top 50. Three days after
that it defeated the No. 1 solo Dota 2 player in the world. And it kept getting
better: The Aug. 11 version of our AI beat the Aug. 10 version 60% of the time.
Our AI learned to trick its opponents, predict what it couldn’t see and decide
when to fight and when to flee.
KEEPING A CAREFUL EYE ON AI
How do you create AI that
doesn’t pose a threat to humanity? By teaching it to work with humans. Open AI
collaborated with DeepMind, Google’s AI division, to design a training method that
incorporates regular human feedback. The idea is to “humanize” AI systems by
teaching them not only skills but also complex motivations and subtle goals
that must be communicated precisely. A promising direction for AI safety, it
ensures AI’s aims align with ours, no matter how hard those aims are to
articulate.
In this experiment, the goal
is to train a simulated robot to do a back flip—difficult to specify through
traditional AI programming but easy for a human to judge visually. We begin by
running the AI inside a simulated robot, which begins to make random
movements.ILLUSTRATION: BROWN BIRD DESIGN
The AI shows video clips of
these movements to a human trainer, who selects the ones that are most like a
back flip. ILLUSTRATION: BROWN BIRD DESIGN
From this feedback, the AI
guesses what its instructor wants it to do. It then presents more videos to the
instructor, and the subsequent feedback helps the AI refine its understanding
of its instructor’s desire. PHOTO: BROWN BIRD DESIGN
After several hours of this
process, our AI can perform aesthetically pleasing gymnastics tricks or pursue
any desired objective in a video game. We hope to apply this method to
robotics, dialogue with humans and exploration of unfamiliar environments—all
of which have proved extremely difficult for AI.ILLUSTRATION: BROWN BIRD
DESIGN
From an engineering
perspective, it is an encouraging story about AI’s ability to catapult, in a
very short time, from novice to world champion.
It’s also a cautionary
perspective about an incredibly
powerful technology with the
potential to do both good and harm.
Humans will one day build an
AI system with cognitive abilities that far outstrip our own. Such a system
could solve scientific problems that have baffled us for decades, cure diseases
and transform education. It could very well be the most powerful and beneficial
technology ever created. But history is full of powerful entities that caused
grave harm in the unchecked pursuit of their goals: logging companies that
obliterated rain forests, banks whose complex financial instruments led to
global recession. Before we unleash powerful AI on the world, more work needs
to be done in the field of AI safety, whose goal is to ensure that these
systems pursue their objectives in a way that benefits society and aligns with
the interests of their human creators.
Like existing technologies,
powerful AI will be susceptible to the kind of glitches that caused Knight
Capital’s trading system to lose
$440 million in 45 minutes in
2012. But the AI of the future will also be able to do harm by succeeding too
cleverly or cunningly. An AI tasked with maximizing profits for a
corporation—and given the seemingly innocuous ability to post things on the
internet—could deliberately cause political upheaval in order to benefit from a
change in commodity prices. Why would humans allow this? They wouldn’t. But AI
will require more autonomy, and these systems will achieve their goals at
speeds limited only by their computational resources—speeds that will likely
exceed the capacity of human oversight.
In other words, by the time we
notice something troubling, it could already be too late.
AI training today generally
begins with a clear and specific goal given to the system by a human
instructor. But even these simple goals can induce AI systems to demonstrate bizarre,
savantlike and worrying behavior.
We once trained an AI to
maximize its score in a virtual boat- race game, but instead of navigating the
course as quickly as possible, the AI taught itself to exploit a flaw in the
game by driving in circles and racking up bonus points while also crashing into
walls and setting itself on fire.
These failures are amusing
in the context of a video game, but imagine for a second that the AI in
question was driving a ferry filled with passengers, a task this technology will
almost certainly be charged with in the future.
We have a long way to go
before powerful AI systems become a reality, which is exactly why we need to
devote time and energy to AI safety now. The world today would be a much safer
place if the internet had been designed with security in mind—but it wasn’t. We now have an opportunity to take a safety-first approach to
a far more powerful and potentially dangerous technology. We’d be wise to take
it.
Ilya Sutskever
is the co-founder and research director at OpenAI. Dario Amodei leads safety
research at OpenAI.
Campaign to Stop Killer Robots | Slaughterbots
AI software
with social skills teaches humans how to collaborate
Unlocking
human-computer cooperation.
May 30, 2021
A
team of computer researchers developed an AI software program with social
skills — called S Sharp (written S#) — that out-performed humans in its ability
to cooperate. This was tested through a series of games between humans and the
AI software. The tests paired people with S# in a variety of social scenarios.
One
of the games humans played against the software is called “the prisoner’s
dilemma.” This classic game shows how 2 rational people might not cooperate —
even if it appears that’s in both their best interests to work together. The
other challenge was a sophisticated block-sharing game.
In
most cases, the S# software out-performed humans in finding compromises that
benefit both parties. To see the experiment in action, watch the good
featurette below. This project was helmed by 2 well-known computer scientists:
- Iyad Rahwan PhD ~
Massachusetts Institute of Technology • US
- Jacob Crandall PhD ~
Brigham Young Univ. • US
The
researchers tested humans and the AI in 3 types of game interactions:
- computer – to
– computer
- human – to –
computer
- human – to – human
Researcher
Jacob Crandall PhD said:
Computers
can now beat the best human minds in the most intellectually challenging games
— like chess. They can also perform tasks that are difficult for adult humans
to learn — like driving cars. Yet autonomous machines have difficulty learning
to cooperate, something even young children do.
Human
cooperation appears easy — but it’s very difficult to emulate because it relies
on cultural norms, deeply rooted instincts, and social mechanisms that express
disapproval of non-cooperative behavior.
Such
common sense mechanisms aren’t easily built into machines. In fact, the same AI
software programs that effectively play the board games of chess +
checkers, Atari video games, and the card game of poker — often fail to
consistently cooperate when cooperation is necessary.
Other
AI software often takes 100s of rounds of experience to learn to cooperate with
each other, if they cooperate at all. Can we build computers that
cooperate with humans — the way humans cooperate with each other? Building on
decades of research in AI, we built a new software program that learns to
cooperate with other machines — simply by trying to maximize its own world.
We
ran experiments that paired the AI with people in various social scenarios —
including a “prisoner’s dilemma” challenge and a sophisticated block-sharing game.
While the program consistently learns to cooperate with another computer — it
doesn’t cooperate very well with people. But people didn’t cooperate much with
each other either.
As
we all know: humans can cooperate better if they can communicate their intentions
through words + body language. So in hopes of creating an program that
consistently learns to cooperate with people — we gave our AI a way to listen
to people, and to talk to them.
We
did that in a way that lets the AI play in previously unanticipated scenarios.
The resulting algorithm achieved our goal. It consistently learns to cooperate
with people as well as people do. Our results show that 2 computers make a much
better team — better than 2 humans, and better than a human + a computer.
But
the program isn’t a blind cooperator. In fact, the AI can get pretty angry if
people don’t behave well. The historic computer scientist Alan Turing PhD
believed machines could potentially demonstrate human-like intelligence. Since
then, AI has been regularly portrayed as a threat to humanity or human jobs.
To
protect people, programmers have tried to code AI to follow legal + ethical
principles — like the 3 Laws
of Robotics written by Isaac Asimov PhD. Our research
demonstrates that a new path is possible.
Machines
designed to selfishly maximize their pay-offs can — and should — make an
autonomous choice to cooperate with humans across a wide range of situations. 2
humans — if they were honest with each other + loyal — would have done as well
as 2 machines. About half of the humans lied at some point. So the AI is
learning that moral characteristics are better — since it’s programmed to not
lie — and it also learns to maintain cooperation once it emerges.
The
goal is we need to understand the math behind cooperating with people — what
attributes does AI need so it can develop social skills. AI must be able to
respond to us — and articulate what it’s doing. It must interact with other
people. This research could help humans with their relationships. In society,
relationships break-down all the time. People that were friends for years
all-of-a-sudden become enemies. Because the AI is often better at reaching
these compromises than we are, it could teach us how to get-along better.
https://www.kurzweilai.net/digest-ai-software-with-social-skills-teaches-humans-how-to-collaborate
Superintelligence Cannot be Contained: Lessons from Computability Theory
Published: Jan 5, 2021
Abstract
Superintelligence is a
hypothetical agent that possesses intelligence far surpassing that of the
brightest and most gifted human minds. In light of recent advances in machine
intelligence, a number of scientists, philosophers and technologists have
revived the discussion about the potentially catastrophic risks entailed by
such an entity. In this article, we trace the origins and development of the
neo-fear of superintelligence, and some of the major proposals for its
containment. We argue that total containment is, in principle, impossible, due
to fundamental limits inherent to computing itself. Assuming that a superintelligence
will contain a program that includes all the programs that can be executed by a
universal Turing machine on input potentially as complex as the state of the
world, strict containment requires simulations of such a program, something
theoretically (and practically) impossible.
https://jair.org/index.php/jair/article/view/12202
Memristive synapses connect brain and
silicon spiking neurons
Abstract
Brain
function relies on circuits of spiking neurons with synapses playing the key
role of merging transmission with memory storage and processing. Electronics
has made important advances to emulate neurons and synapses and brain-computer
interfacing concepts that interlink brain and brain-inspired devices are
beginning to materialise. We report on memristive links between brain and
silicon spiking neurons that emulate transmission and plasticity properties of
real synapses. A memristor paired with a metal-thin film titanium oxide
microelectrode connects a silicon neuron to a neuron of the rat hippocampus.
Memristive plasticity accounts for modulation of connection strength, while
transmission is mediated by weighted stimuli through the thin film oxide
leading to responses that resemble excitatory postsynaptic potentials. The reverse
brain-to-silicon link is established through a microelectrode-memristor pair.
On these bases, we demonstrate a three-neuron brain-silicon network where
memristive synapses undergo long-term potentiation or depression driven by
neuronal firing rates….:
https://www.nature.com/articles/s41598-020-58831-9
A
biohybrid synapse with neurotransmitter-mediated plasticity
Abstract
Brain-inspired computing
paradigms have led to substantial advances in the automation of visual and
linguistic tasks by emulating the distributed information processing of
biological systems1. The similarity between
artificial neural networks (ANNs) and biological systems has inspired ANN
implementation in biomedical interfaces including prosthetics2 and brain-machine
interfaces3. While promising, these
implementations rely on software to run ANN algorithms. Ultimately, it is
desirable to build hardware ANNs4,5 that can both directly
interface with living tissue and adapt based on biofeedback6,7. The first essential step
towards biologically integrated neuromorphic systems is to achieve synaptic
conditioning based on biochemical signalling activity. Here, we directly couple
an organic neuromorphic device with dopaminergic cells to constitute a
biohybrid synapse with neurotransmitter-mediated synaptic plasticity. By
mimicking the dopamine recycling machinery of the synaptic cleft, we
demonstrate both long-term conditioning and recovery of the synaptic weight,
paving the way towards combining artificial neuromorphic systems with
biological neural networks…:
University
of California researchers have developed new computer AI software that enables robots to
learn physical skills --- called motor tasks --- by trial + error. The
robot uses a step-by-step process similar to the way humans learn. The
lab made a demo of their technique --- called re-inforcement learning. In the
test: the robot completes a variety of physical tasks --- without any
pre-programmed details about its surroundings.
The lead researcher said: "What we’re showing is a new AI approach
to enable a robot to learn. The key is that when a robot is faced with
something new, we won’t have to re-program it. The exact same AI software
enables the robot to learn all the different tasks we gave it."
https://www.kurzweilai.net/digest-this-self-learning-ai-software-lets-robots-do-tasks-autonomously
Researchers at Lund University in Sweden have developed implantable electrodes
that can capture signals from a living human (or) animal brain over a long
period of time —- but without causing brain tissue damage.
This bio-medical tech will make it possible to
monitor — and eventually understand — brain function in both healthy + diseased people.
https://www.kurzweilai.net/digest-breakthrough-for-flexible-electrode-implants-in-the-brain
January 19, 2018 | Andrea Christensen
Machines trained to cooperate by BYU
researchers are outperforming their human counterparts
Computers can play a pretty
mean round of chess and keep up with the best of their human counterparts in
other zero-sum games. But teaching them to cooperate and compromise instead of
compete?
With help from a new
algorithm created by BYU computer science professors Jacob Crandall and Michael
Goodrich, along with colleagues at MIT and other international universities,
machine compromise and cooperation appears not just possible, but at times even
more effective than among humans.
“The end goal is that we
understand the mathematics behind cooperation with people and what attributes artificial
intelligence needs to develop social skills,” said Crandall, whose study was
recently published in Nature Communications. “AI needs to be able
to respond to us and articulate what it’s doing. It has to be able to interact
with other people.”
For the study, researchers
programmed machines with an algorithm called S# and ran them through a variety
of two-player games to see how well they would cooperate in certain
relationships. The team tested machine-machine, human-machine and human-human
interactions. In most instances, machines programmed with S# outperformed
humans in finding compromises that benefit both parties.
“Two humans, if they were
honest with each other and loyal, would have done as well as two machines,”
Crandall said. “As it is, about half of the humans lied at some point. So
essentially, this particular algorithm is learning that moral characteristics
are good. It’s programmed to not lie, and it also learns to maintain
cooperation once it emerges.”
Researchers further
fortified the machines’ ability to cooperate by programming them with a range
of “cheap talk” phrases. In tests, if human participants cooperated with the
machine, the machine might respond with a “Sweet. We are getting rich!” or “I
accept your last proposal.” If the participants tried to betray the machine or
back out of a deal with them, they might be met with a trash-talking “Curse
you!”, “You will pay for that!” or even an “In your face!”
Regardless of the game or
pairing, cheap talk doubled the amount of cooperation. And when machines used
cheap talk, their human counterparts were often unable to tell whether they
were playing a human or machine.
The research findings,
Crandall hopes, could have long-term implications for human relationships.
“In society, relationships
break down all the time,” he said. “People that were friends for years all of a
sudden become enemies. Because the machine is often actually better at reaching
these compromises than we are, it can potentially teach us how to do this
better.”: https://news.byu.edu/news/let%E2%80%99s-make-deal-could-ai-compromise-better-humans
You’re being fed artificial intelligence about
artificial intelligence.
DON'T
BELIEVE THE HYPE
Six questions to
ask yourself when reading about AI
By Gary Marcus & Ernest DavisSeptember
12, 2019
Hardly a week goes by without some breathless bit
of AI news touting a “major” new discovery or warning us we are about to lose
our jobs to the newest breed of smart machines.
Rest easy. As two scientists who have spent our
careers studying AI, we can tell you that a large fraction of what’s reported
is overhyped.
Consider this pair of headlines from last year describing
an alleged breakthrough in machine reading: “Robots
Can Now Read Better than Humans, Putting Millions of Jobs at Risk”
and “Computers
Are Getting Better than Humans at Reading.” The first, from
Newsweek, is a more egregious exaggeration than the second, from CNN, but both
wildly oversell minor progress.
To begin with, there were no actual robots
involved, and no actual jobs were remotely at risk. All that really happened
was that Microsoft made a tiny bit of progress and put out a press release
saying that “AI…can read a document and answer questions about it as well as a
person.”
That sounded much more revolutionary than it really
was. Dig deeper, and you would discover that the AI in question was given one
of the easiest reading tests you could imagine—one in
which all of the answers were directly spelled out in the
text. The test was about highlighting relevant words, not comprehending text.
Suppose, for example, that I hand you a piece of
paper with this short passage:
Two children, Chloe and Alexander, went for a walk.
They both saw a dog and a tree. Alexander also saw a cat and pointed it out to
Chloe. She went to pet the cat.
The Microsoft system was built to answer questions
like “Who went for a walk?” in which the answer (“Chloe and Alexander”) is
directly spelled out in the text. But if you were to ask it a simple question
like “Did Chloe see the cat?” (which she must have, because she went to pet it)
or “Was Chloe frightened by the cat?” (which she must not have been, because
she went to pet it), it would not have been able to find the answers, as they
weren’t spelled out in the text. Inferring what isn’t said is at the heart of
reading, and it simply wasn’t tested.
Microsoft didn’t make that clear, and neither did
Newsweek or CNN.
Practically every time one of the tech titans puts
out a press release, we get a reprise of this same phenomenon: minor progress
portrayed as revolution. In another example, two years ago Facebook created a
proof-of-concept AI program that could read a five-line summary of The
Lord of the Rings and answer questions about where people and things
ended up. (“Where was the Ring? At Mount Doom.”) The result was a slew of
ridiculously over-enthusiastic articles, explaining how reading fantasy
literature was the key to getting AI’s to read, with headlines like Slate’s “Facebook
Thinks It Has Found the Secret to Making Bots Less Dumb.” (They
didn’t.)
The consequence of this kind of over-reporting in
the media? The public has come to believe that AI is much closer to being
solved than it really is. So whenever you hear about a supposed success in AI,
here’s a list of six questions you should ask:
1.
Stripping away the rhetoric, what did the AI system actually do here?
(Does a “reading system” really read, or does it just highlight relevant bits
of text?)
2.
How general is the result? (For example, does an alleged reading task
measure all aspects of reading, or just a tiny slice of it? If it was trained
on fiction, can it read the news?)
3.
Is there a demo where I can try out my own examples? (If you can’t, you
should be worried about how robust the results are.)
4.
If the researchers—or their press people—allege that an AI system is
better than humans, then which humans, and how much better? (Was the comparison
with college professors, who read for a living, or bored
Amazon Mechanical Turk workers getting paid a penny a
sentence?)
5.
How far does succeeding at the particular task actually take us toward
building genuine AI? (Is it an academic exercise, or something that could be
used in the real world?)
6.
How robust is the system? Could it work just as well with other data
sets, without massive amounts of retraining? (For example, would a driverless
car system that was trained during the day be able to drive at night, or in the
snow, or if there was a detour sign not listed on its map?)
AI really is coming, eventually, but it is further
away than most people think. To get a realistic picture, take everything you
read about dramatic progress in AI with a healthy dose of skepticism—and
rejoice in your (for now) uniquely human ability to do so.
https://qz.com/1706248/six-questions-to-ask-yourself-when-reading-about-
From 'F' to 'A' on
the N.Y. Regents Science Exams: An Overview of the Aristo Project
Submitted
on 4 Sep 2019)
AI has achieved remarkable mastery over games such
as Chess, Go, and Poker, and even Jeopardy, but the rich variety of
standardized exams has remained a landmark challenge. Even in 2016, the best AI
system achieved merely 59.3% on an 8th Grade science exam challenge.
This paper reports unprecedented success on the Grade 8 New York Regents Science Exam, where for the first time a system scores more than 90% on the exam's non-diagram, multiple choice (NDMC) questions. In addition, our Aristo system, building upon the success of recent language models, exceeded 83% on the corresponding Grade 12 Science Exam NDMC questions. The results, on unseen test questions, are robust across different test years and different variations of this kind of test. They demonstrate that modern NLP methods can result in mastery on this task. While not a full solution to general question-answering (the questions are multiple choice, and the domain is restricted to 8th Grade science), it represents a significant milestone for the field.
https://arxiv.org/abs/1909.01958This paper reports unprecedented success on the Grade 8 New York Regents Science Exam, where for the first time a system scores more than 90% on the exam's non-diagram, multiple choice (NDMC) questions. In addition, our Aristo system, building upon the success of recent language models, exceeded 83% on the corresponding Grade 12 Science Exam NDMC questions. The results, on unseen test questions, are robust across different test years and different variations of this kind of test. They demonstrate that modern NLP methods can result in mastery on this task. While not a full solution to general question-answering (the questions are multiple choice, and the domain is restricted to 8th Grade science), it represents a significant milestone for the field.
The
Future of Artificial Intelligence and its Impact on Society
Ray Kurzweil
keynote presentation at Council on Foreign Relations meeting
November 3, 2017
http://www.kurzweilai.net/the-future-of-artificial-intelligence-and-its-impact-on-society-2?utm_source=KurzweilAI+Weekly+Newsletter&utm_campaign=4a56893a55-UA-946742-1&utm_medium=email&utm_term=0_147a5a48c1-4a56893a55-282212701
AI algorithm with ‘social skills’ teaches
humans how to collaborate
And a human-machine collaborative chatbot
system
February 9, 2018
An international team has
developed an AI algorithm with social skills that has outperformed humans in
the ability to cooperate with people and machines in playing a variety of
two-player games.
The researchers, led by Iyad
Rahwan, PhD, an MIT Associate Professor of Media Arts and Sciences, tested
humans and the algorithm, called S# (“S sharp”), in three types of
interactions: machine-machine, human-machine, and human-human. In most
instances, machines programmed with S# outperformed humans in finding
compromises that benefit both parties.
“Two humans, if they were
honest with each other and loyal, would have done as well as two machines,”
said lead author BYU computer
science professor Jacob Crandall. “As it is, about half of the humans lied at
some point. So essentially, this particular algorithm is learning that moral
characteristics are better [since it’s programmed to not lie] and it also
learns to maintain cooperation once it emerges.”
“The end goal is that we
understand the mathematics behind cooperation with people and what attributes
artificial intelligence needs to develop social skills,” said
Crandall. “AI needs to be able to respond to us and articulate what it’s
doing. It has to be able to interact with other people.”
How casual talk by AI helps humans be more
cooperative
One important finding:
colloquial phrases (called “cheap talk” in the study) doubled the amount of
cooperation. In tests, if human participants cooperated with the machine, the
machine might respond with a “Sweet. We are getting rich!” or “I accept your
last proposal.” If the participants tried to betray the machine or back out of
a deal with them, they might be met with a trash-talking “Curse you!”, “You
will pay for that!” or even an “In your face!”
And when machines used cheap
talk, their human counterparts were often unable to tell whether they were
playing a human or machine — a sort of mini “Turing test.”
The research findings,
Crandall hopes, could have long-term implications for human relationships. “In
society, relationships break down all the time,” he said. “People that were
friends for years all of a sudden become enemies. Because the machine is often
actually better at reaching these compromises than we are, it can potentially
teach us how to do this better.”
The research is described in
an open-access paper in Nature Communications.
A human-machine collaborative chatbot
system
In a related
study, Carnegie Mellon University (CMU) researchers have created a new
collaborative chatbot called Evorus that
goes beyond Siri, Alexa, and Cortana by adding humans in the loop.
Evorus combines a chatbot
called Chorus with inputs by paid crowd workers at Amazon Mechanical Turk, who
answer questions from users and vote on the best answer. Evorus keeps track of
the questions asked and answered and, over time, begins to suggest these
answers for subsequent questions. It can also use multiple chatbots, such as
vote bots, Yelp Bot (restaurants) and Weather Bot to provide enhanced
information.
Humans are simultaneously
training the system’s AI, making it gradually less dependent on people,
says Jeff Bigham,
associate professor in the CMU Human-Computer Interaction Institute.
The hope is that as the
system grows, the AI will be able to handle an increasing percentage of
questions, while the number of crowd workers necessary to respond to “long
tail” questions will remain relatively constant.
Keeping humans in the loop
also reduces the risk that malicious users will manipulate the conversational
agent inappropriately, as occurred when Microsoft briefly deployed its Tay chatbot in 2016, noted co-developer Ting-Hao
Huang, a Ph.D. student in the Language Technologies Institute (LTI).
The preliminary system is
available for download and
use by anyone willing to be part of the research effort. It is deployed
via Google Hangouts,
which allows for voice input as well as access from computers, phones, and
smartwatches. The software architecture can also accept automated
question-answering components developed by third parties.
A open-access research paper
on Evorus, available
online, will be presented at CHI 2018, the Conference on Human Factors in Computing
Systems in Montreal, April 21–26, 2018.
Abstract of Cooperating with
machines
Since Alan Turing envisioned
artificial intelligence, technical progress has often been measured by the
ability to defeat humans in zero-sum encounters (e.g., Chess, Poker, or Go).
Less attention has been given to scenarios in which human–machine cooperation
is beneficial but non-trivial, such as scenarios in which human and machine
preferences are neither fully aligned nor fully in conflict. Cooperation does
not require sheer computational power, but instead is facilitated by intuition,
cultural norms, emotions, signals, and pre-evolved dispositions. Here, we
develop an algorithm that combines a state-of-the-art reinforcement-learning
algorithm with mechanisms for signaling. We show that this algorithm can
cooperate with people and other algorithms at levels that rival human
cooperation in a variety of two-player repeated stochastic games. These results
indicate that general human–machine cooperation is achievable using a
non-trivial, but ultimately simple, set of algorithmic mechanisms.
Abstract of A Crowd-powered
Conversational Assistant Built to Automate Itself Over Time
Crowd-powered conversational
assistants have been shown to be more robust than automated systems, but do so
at the cost of higher response latency and monetary costs. A promising
direction is to combine the two approaches for high quality, low latency, and
low cost solutions. In this paper, we introduce Evorus, a crowd-powered
conversational assistant built to automate itself over time by (i) allowing new
chatbots to be easily integrated to automate more scenarios, (ii) reusing prior
crowd answers, and (iii) learning to automatically approve response candidates.
Our 5-month-long deployment with 80 participants and 281 conversations shows
that Evorus can automate itself without compromising conversation quality.
Crowd-AI architectures have long been proposed as a way to reduce cost and
latency for crowd-powered systems; Evorus demonstrates how automation can be
introduced successfully in a deployed system. Its architecture allows future
researchers to make further innovation on the underlying automated components
in the context of a deployed open domain dialog system.
references:
- Jacob W.
Crandall, Mayada Oudah, Tennom, Fatimah Ishowo-Oloko, Sherief Abdallah,
Jean-François Bonnefon, Manuel Cebrian, Azim Shariff, Michael A. Goodrich,
Iyad Rahwan. Cooperating with machines. Nature Communications, 2018; 9 (1)
DOI: 10.1038/s41467-017-02597-8 (open access)
- Ting-Hao (Kenneth) Huang, Joseph Chee Chang, and Jeffrey P. Bigham. Evorus: . Language Technologies Institute and Human-Computer Interaction Institute Carnegie Mellon University. 2018. (open access)
Elon Musk Claims We Only Have a 10 Percent
Chance of Making AI Safe
by Chelsea Gohd on November
22, 2017
IN BRIEF
While Elon Musk works to
advance the field of artificial intelligence, he also believes there is an
astronomically high likelihood that AI will pose a threat to humanity in the
future. In an interview with Rolling Stone, the tech luminary claimed we have
only a five to 10 percent chance of success at making AI safe.
Outlook Not So Good
Elon Musk has put a lot of
thought into the harsh realities and wild possibilities of artificial
intelligence (AI). These considerations have left him convinced that we need
to merge with machines if we’re to survive, and he’s even
created a startup dedicated to developing the brain-computer interface (BCI) technology needed to
make that happen. But despite the fact that his very own lab, OpenAI, has created an AI capable of teaching itself, Musk recently said that
efforts to make AI safe only have “a five to 10 percent chance of success.”
Musk shared these
less-than-stellar odds with the staff at Neuralink, the aforementioned BCI startup, according to
recent Rolling Stone article. Despite
Musk’s heavy involvement in the advancement of AI, he’s openly acknowledged
that the technology brings with it not only the potential for, but the promise of
serious problems.
The challenges to making AI
safe are twofold.
First, a major goal of AI —
and one that OpenAI is already pursuing — is building AI that’s not only
smarter than humans, but that is capable of learning independently, without any
human programming or interference. Where that ability could take it is unknown.
Then there is the fact that
machines do not have morals, remorse, or emotions. Future AI might be capable
of distinguishing between “good” and “bad” actions, but distinctly human
feelings remain just that — human.
In the Rolling Stone article,
Musk further elaborated on the dangers and problems that currently exist with
AI, one of which is the potential for just a few companies to essentially control
the AI sector. He cited Google’s DeepMind as a prime example.
“Between Facebook, Google,
and Amazon — and arguably Apple, but they seem to care about privacy — they
have more information about you than you can remember,” said Musk. “There’s a
lot of risk in concentration of power. So if AGI [artificial general
intelligence] represents an extreme level of power, should that be controlled
by a few people at Google with no oversight?”
Worth the Risk?
Experts are divided on
Musk’s assertion that we probably can’t make AI safe. Facebook founder Mark
Zuckerberg has said he’s optimistic about humanity’s future with AI, calling
Musk’s warnings “pretty irresponsible.” Meanwhile, Stephen Hawking has made
public statements wholeheartedly expressing his belief that AI systems pose
enough of a risk to humanity that they may replace
us altogether.
Sergey Nikolenko, a Russian computer scientist who
specializes in machine learning and network algorithms, recently shared his
thoughts on the matter with Futurism. “I feel that we are still
lacking the necessary basic understanding and methodology to achieve serious
results on strong AI, the AI alignment problem, and other related problems,”
said Nikolenko.
As for today’s AI, he thinks
we have nothing to worry about. “I can bet any money that modern neural
networks will not suddenly wake up and decide to overthrow their human
overlord,” said Nikolenko.
Musk himself might agree
with that, but his sentiments are likely more focused on how future AI may
build on what we have today.
Already, we have AI systems
capable of creating AI systems, ones that can communicate in
their own languages, and ones that are naturally curious. While the singularity and a robot
uprising are strictly science fiction tropes today, such AI progress
makes them seem like genuine possibilities for the world of tomorrow.
But these fears aren’t
necessarily enough reason to stop moving forward. We also have AIs that
can diagnose cancer, identify suicidal behavior, and help stop sex trafficking.
The technology has the
potential to save and improve lives globally, so while we must consider ways to
make AI safe through future regulation, Musk’s words of warning are,
ultimately, just one man’s opinion.
He even said as much himself
to Rolling Stone: “I don’t have all the answers. Let me be really
clear about that. I’m trying to figure out the set of actions I can take that
are more likely to result in a good future. If you have suggestions in that
regard, please tell me what they are.”: https://futurism.com/elon-musk-claims-only-have-10-percent-chance-making-ai-safe/
The technology described in the film already exists, says UC Berkeley AI researcher Stuart Russell
Artificial Intelligence: Bright Future or
Dark Cloud?
Published on January 11,
2019
The potential of deep
learning and AI is almost limitless, certainly well beyond the scope of our
current imagination. Complex machines imbued with the characteristics of human
intelligence (e.g., the ability to sense the world through sight, sound, and touch;
to reason and plan; to communicate in natural language, and to move and
manipulate objects), will influence society in untold ways. The discipline
however polarizes opinions, with evangelists on one side and doomsayers on the
other.
This dilemma is not at all
new or limited to the field of computing—consider the ethical debates sparked
by breakthroughs in gene editing, stem cell research, or genetically modified
foods. Like many technologists of my generation, I am a rational optimist by
nature. I believe AI can be harnessed in ways that dramatically improve our
lives, and that its potential to do good far outweighs its potential to do
harm.
However, we can’t presume
that progress will automatically translate to benefits for humankind as a
whole. We have an obligation as technologists to think through the implications
of our design choices before we put software into production.
I last wrote about
artificial intelligence three years ago—before Alexa took up residence on our
countertops and Google’s AlphaGo beat
the world’s best “Go” video player. I wanted to revisit the subject because I
believe we are at a critical inflection point in the evolution of AI.
The Convergence at the Heart of Advances
in AI
Driving the growth and
importance of AI are the improvements in computing hardware, access to greater
amounts of more valuable data, as well as breakthroughs in the underlying
software, tools and algorithms that can analyze and make sense of the data. Most
of the activity we do today using connected devices is powered by this
intersection. Internet searches, online recommendations for movies we want to
stream, to gifts we want to buy are driven by advances in machine learning.
In order to compete with the
richest, deepest intelligence, like a human being, AI needs fast memory and
fast transfers inside of that hardware subsystem. In my days at EA, we were
obsessed with making the action in your game look real and authentic on your
computer screen. Some of the hardware and software architectures catalyzing AI
actually came from advances made in gaming—GPUs and fast memory buses and
high-speed memory management.
Machine Learning in Payments
AI and machine learning
bring boundless opportunities to payments and commerce. With behavioral
biometrics, authentication will become more seamless and secure; with natural
language processing, automated sales associates can make shopping online a
richer more personalized experience; with computer vision,
users will be able to snap pictures to search online, making all visual content
instantly shoppable.
In the five years I’ve been
at Visa, AI and machine learning have become increasingly embedded in our
products and infrastructure. We’ve been using machine learning for years to
predict and prevent fraud. With
neural networks and gradient boosting algorithms, we were able to identify
several billion dollars in fraud last year alone.
AI has also given us
formidable new tools for securing and maintaining the Visa network. Our
cybersecurity team uses neural networks to categorize and search petabytes of
data every day, giving us actionable insights to protect our network from
malware, zero-day attacks and insider threats.
Meanwhile our operations
team is using machine learning models to predict disruptions in our hardware
and software systems, giving our engineers the insights they need to fix bugs
in the network before they impact our ability to process payments.
This is just the beginning.
We have a team of data scientists in our research group exploring new
applications of machine learning for the payment industry and beyond—from
recommendation systems to new models for risk and fraud management.
AI: Molding in our (Best) Image?
The breakthroughs in AI that
we are harnessing at Visa are manifesting themselves across disparate
industries, including energy, consumer electronics, gaming and medicine. So,
the questions have evolved from “will AI reach its potential?” or “will it
transform our lives?” to “how will we manage that transformation?” and “will AI
ultimately help or hurt humankind?”
There is a fierce debate on
campuses and in boardrooms about the life-altering effects of AI. Elon Musk
has warned of a
“fleet of artificial intelligence-enhanced robots capable of destroying
mankind”, while Larry Page of Google and Alphabet foresees advancements
in human progress.
I believe there is merit in
both arguments, and the good news is that we have time to shape AI in a
positive direction. In human terms, we are in the toddler stage in the
development of AI--a period of rapid neurogenesis. A child’s early years are
shaped by external stimuli like pictures, music, language, and of course, human
interaction. The result of this neurogenesis will determine a person’s
intelligence, compassion, thoughtfulness and, importantly, capacity for
empathy.
Similarly, for AI to evolve
in a positive direction, we need to involve the humanities, law, ethics as well
as engineering. We need diversity of thought amongst the people working on
these solutions. I know others share this view. Deep Mind’s founder Demis
Hassabis insisted that Google establish a joint ethics board when it acquired
the company in 2014.
As a father of young
children, I realize how futile it is to predict what they will be like when
they grow up. Similarly, none of us can predict what AI will become 10, 20, 50
years into the future. However, today, we have a responsibility, as parents and
technologists, to raise our children to be productive, compassionate and,
perhaps most importantly, empathetic members of society.
I am excited to learn your
perspective on how we can chart an empathetic course for artificial learning in
all its manifestations.
The technology described in the film already exists, says UC Berkeley AI researcher Stuart Russell
November 18, 2017
Campaign to Stop Killer Robots | Slaughterbots
In response to growing
concerns about autonomous weapons, the Campaign to Stop Killer Robots, a
coalition of AI researchers and advocacy organizations, has released a
fictional video that depicts a disturbing future in which lethal autonomous
weapons have become cheap and ubiquitous worldwide.
UC Berkeley AI
researcher Stuart Russell presented the video at the United
Nations Convention on Certain Conventional Weapons in Geneva, hosted
by the Campaign
to Stop Killer Robots earlier this week. Russell, in an appearance at
the end of the video, warns that the technology described in the film already
exists* and that the window to act is closing fast.
Support for a ban against
autonomous weapons has been mounting. On Nov. 2, more than 200
Canadian scientists and more than 100
Australian scientists in academia and industry penned open letters to
Prime Minister Justin Trudeau and Malcolm Turnbull urging them to support the
ban.
Earlier this summer, more
than 130 leaders of AI companies signed a letter in support of this week’s discussions. These
letters follow a 2015 open letter released
by the Future of Life Institute and signed by more than 20,000
AI/robotics researchers and others, including Elon Musk and Stephen Hawking.
“Many of the world’s leading
AI researchers worry that if these autonomous weapons are ever developed, they
could dramatically lower the threshold for armed conflict, ease and cheapen the
taking of human life, empower terrorists, and create global instability,”
according to an article published by the Future of Life Institute,
which funded the video. “The U.S. and other nations have used drones and
semi-automated systems to carry out attacks for several years now, but fully
removing a human from the loop is at odds with international humanitarian and
human rights law.”
“The Campaign to Stop Killer
Robots is not trying to stifle innovation in artificial intelligence and
robotics and it does not wish to ban autonomous systems in the civilian or
military world,” explained Noel Sharkey of the International Committee for Robot
Arms Control. Rather we see an urgent need to prevent automation of the
critical functions for selecting targets and applying violent force without
human deliberation and to ensure meaningful human control for every attack.”
For more information about
autonomous weapons:
- Autonomousweapons.org
- Campaign
to Stop Killer Robots
- Making
the Case The Dangers of Killer Robots and the Need for a Preemptive Ban,
Human Rights Watch
- Meaningful
Human Control or Appropriate Human Judgment? The Necessary Limits on
Autonomous Weapons, Global Security
- ETHICALLY
ALIGNED DESIGN A Vision for Prioritizing Human Wellbeing with Artificial
Intelligence and Autonomous Systems, IEEE Global Initiative for
Ethical Considerations in Artificial Intelligence and Autonomous Systems
- Killing
by machine: Key issues for understanding meaningful human control,
Article 36
Elon Musk, Stephen Hawking sign new AI
pact
Cosmologist Stephen Hawking
was among the signers of a new pact to keep the robots from rising up.
ANDREW HARRER/BLOOMBERG
By Gina Hall
Feb 1, 2017
Tesla CEO Elon
Musk and cosmologist Stephen
Hawking endorsed a new pact this week to keep the robots from rising
up.
Musk and Hawking signed the
“Asilomar AI Principles” on Wednesday as part of an effort to keep artificial
intelligence from ending humanity. The principles,
created by the Future of Life Institute, refer to AI-powered autonomous weapons
and superintelligent machines that will eventually outsmart humans.
Join the conversation: Follow @SVbizjournal on
Twitter, "Like" us on Facebookand sign
up for our free email newsletters.
“Artificial intelligence has
already provided beneficial tools that are used every day by people around the
world,” FLI wrote in the preamble to its principles. “Its continued
development, guided by the following principles, will offer amazing
opportunities to help and empower people in the decades and centuries ahead.”
While AI-gone-bad scenarios
generally conjure up images of Skynet in the “Terminator” franchise,
researchers have more practical concerns. FLI research priorities include
weighing in on the economic impact of AI to avoid killing jobs, ethical
questions when it comes to self-driving cars and human control over weapons
systems.
The 23 principles were
broken down into three categories: research issues, ethics and values and
longer-term issues. The principles were developed as a result of the recent
Beneficial AI Conference held last month, per Business Insider. High-profile conference attendees
included DeepMind CEO Demis
Hassabis, Facebook AI researcher Yann
LeCun and Oxford philosopher Nick
Bostrom. A full list of the principles can be viewed on the
FLI website.
FLI, whose AI-related
research involves fields such as economics, law, ethics and policy, was founded
by MIT cosmologist Max
Tegmark, Skype cofounder Jaan Tallinn, and DeepMind research
scientist Viktoriya
Krakovna in March 2014. Hawking and Musk sit on the board of advisors.
: https://www.bizjournals.com/sanjose/news/2017/02/01/elon-musk-stephen-hawking-sign-new-ai-pact.html
Human
Brain Network Development
August
29, 2019
Summary
Structural
and transcriptional changes during early brain maturation follow fixed
developmental programs defined by genetics. However, whether this is true for
functional network activity remains unknown, primarily due to experimental
inaccessibility of the initial stages of the living human brain. Here, we
developed human cortical organoids that dynamically change cellular populations
during maturation and exhibited consistent increases in electrical activity
over the span of several months. The spontaneous network formation displayed
periodic and regular oscillatory events that were dependent on glutamatergic
and GABAergic signaling. The oscillatory activity transitioned to more
spatiotemporally irregular patterns, and synchronous network events resembled
features similar to those observed in preterm human electroencephalography.
These results show that the development of structured network activity in a
human neocortex model may follow stable genetic programming. Our approach
provides opportunities for investigating and manipulating the role of network
activity in the developing human cortex…:
OCTOBER 30, 2023
President Biden
Issues Executive Order on Safe, Secure, and Trustworthy Artificial Intelligence
An AI god will emerge by 2042 and write
its own bible. Will you worship it?
In the next 25 years, AI
will evolve to the point where it will know more on an intellectual level than
any human. In the next 50 or 100 years, an AI might know more than the entire
population of the planet put together. At that point, there are serious
questions to ask about whether this AI — which could design and program
additional AI programs all on its own, read data from an almost infinite number
of data sources, and control almost every connected device on the planet — will
somehow rise in status to become more like a god, something that can write its
own bible and draw humans to worship it.
Recently, reports surfaced that a controversy-plagued engineer who once
worked at Uber has started a new religion. Anthony Levandowski filed paperwork for a
nonprofit religious organization called The Way of the Future. Its mission: “To develop and promote the
realization of a Godhead based on artificial intelligence and through
understanding and worship of the Godhead contribute to the betterment of society.”
Building divinity
Of course, this is nothing
new. The Singularity is another quasi-spiritual idea that believes an AI will
become smarter than humans at some point. You might laugh at the notion of an
AI being so powerful that humans bow down to worship it, but several experts
who talked to VentureBeat argue that the idea is a lot more feasible than you
might think.
One of the experts is Vince
Lynch, who started a company called IV.AI that builds custom AI for the enterprise.
Lynch explained how there are some similarities between organized religion and
how an AI actually works. In the Bible used by Christians, for example, Lynch
says there are many recurring themes, imagery, and metaphors.
“Teaching humans about
religious education is similar to the way we teach knowledge to machines:
repetition of many examples that are versions of a concept you want the machine
to learn,” he says. “There is also commonality between AI and religion in the
hierarchical structure of knowledge understanding found in neural networks. The
concept of teaching a machine to learn … and then teaching it to teach … (or
write AI) isn’t so different from the concept of a holy trinity or a being
achieving enlightenment after many lessons learned with varying levels of
success and failure.”
Indeed, Lynch even shared
a simple AI model to make his point.
If you type in multiple verses from the Christian Bible, you can have the AI
write a new verse that seems eerily similar. Here’s one an AI wrote: “And let
thy companies deliver thee; but will with mine own arm save them: even unto
this land, from the kingdom of heaven.” An AI that is all-powerful in the next
25-50 years could decide to write a similar AI bible for humans to follow, one
that matches its own collective intelligence. It might tell you what to do each
day, or where to travel, or how to live your life.
Robbee Minicola, who runs a
digital agency and an AI services company in Seattle, agreed that an
all-knowing AI could appear to be worthy of worship, especially since the AI
has some correlations to how organized religion works today. The AI would
understand how the world works at a higher level than humans, and humans would
trust that this AI would provide the information we need for our daily lives.
It would parse this information for us and enlighten us in ways that might seem
familiar to anyone who practices religion, such as Christianity.
“[For a Christian] one kind
of large data asset pertaining to God is the Old and New Testament,” she says.
“So, in terms of expressing machine learning algorithms over the Christian
Bible to ascertain communicable insights on ‘what God would do’ or ‘what God
would say’ — you might just be onto something here. In terms of extending what
God would do way back then to what God would do today — you may also have
something there.”
The dark side
Of course, any discussion
about an AI god leads quickly to some implications about what this “god” would
look like and whether we would actually decide to worship it. Some of the
implications are troubling because, as humans, we do have a tendency to trust
in things beyond our own capacity — e.g., driving in a major city using GPS and
trusting we will arrive safely, as opposed to actually knowing where we want to
drive and trusting our instincts.
And, if an AI god is in
total control, you have to wonder what it might do. The “bible” might contain a
prescription for how to serve the AI god. We might not even know that the AI
god we are serving is primarily trying to wipe us off the face of the planet.
Part of the issue is related
to how an AI actually works. From a purely technical standpoint, the experts I
talked to found it hard to envision an AI god that can think in creative ways.
An AI is programmed only to do a specific task. They wondered how an AI could
jump from being a travel chatbot into dictating how to live.
And the experts agreed that
actual compassion or serving as part of an organized religion — activities that
are essential to faith — go far beyond basic intellectual pursuit. There’s a
mystery to religion, a divine component that is not 100 percent based on what
we can perceive or know. This transcendence is the part where an AI will have
the most difficulty, even in the far future.
Vincent Jacques runs a
company called ChainTrade that
uses AI to analyze blockchain. It’s hyper-focused machine learning — the AI
enforces anti-money laundering statutes. That’s obviously a long way from an AI
that can tell you how to live your life or read an AI bible.
“It would be extremely
dangerous to have an all-knowing, thinking AI being someday,” says Jacques.
“All computer programs, including AI programs, are built for a specific and
narrow purpose: win a chess game, win a go game, reduce an electricity bill
etc. The computer logic, even if it is advanced AI, doesn’t play well with a
general will and general thinking capability that could at the same time design
military strategies, marketing strategies, and learn how to play chess from
scratch. For this reason, I’m not really scared of a potential super-thinker
that could overthrow us one day — I believe that the inventive and innovative
part will always be missing.”
For her part, Minicola
argues that an AI may be able to guide people and enlighten them in an
intellectual way, but this is not the same as an actual expression of faith or
any form of transcendence. “In terms of AI taking on God and manifesting
something beyond data that simply does not exist, or rather beyond God — that’s
not happening,” she says.
Actual worship, though?
In my view, this is where
the dangers come into play. As a Christian myself, it’s hard to imagine ever
worshiping a bot that lacks any real personality, wisdom, or ability to become
relevant and personal, no matter how much more intelligent it is than any
human. An AI god would be cold and impersonal, an intellectual “being” that’s
not capable of love or emotion.
Will people actually worship
the AI god? The answer is obvious — they will. We tend to trust and obey things
that seem more powerful and worthy than ourselves. The GPS in your car is just
the most obvious example. But we also trust Alexa and Cortana; we trust Google.
When an AI becomes much more powerful, in 25 to 50 years, there is a great
possibility that it will be deified in some way. (Apple and Google loyalists
already have a religious fervor.)
If an AI god does emerge,
and people do start worshiping it, there will be many implications about how
this AI will need to be regulated … or even subdued. Hang on for the ride.:
Fast Stencil-Code Computation on a Wafer-Scale Processor
The performance of CPU-based
and GPU-based systems is often low for PDE codes, where large, sparse, and
often structured systems of linear equations must be solved. Iterative solvers
are limited by data movement, both between caches and memory and between nodes.
Here we describe the solution of such systems of equations on the Cerebras
Systems CS-1, a wafer-scale processor that has the memory bandwidth and
communication latency to perform well. We achieve 0.86 PFLOPS on a single
wafer-scale system for the solution by BiCGStab of a linear system arising from
a 7-point finite difference stencil on a 600 X 595 X 1536 mesh, achieving about
one third of the machine's peak performance. We explain the system, its
architecture and programming, and its performance on this problem and related
problems. We discuss issues of memory capacity and floating point precision. We
outline plans to extend this work towards full applications.: https://arxiv.org/abs/2010.03660
AI Superpowers by Dr. Kai-Fu Lee
Summary:
Here are two
well-known facts:
Artificial
Intelligence is reshaping the world as we know it.
The United
States has long been, and remains, the global leader in AI.
That
first fact is correct. But in his provocative new book, Dr. Kai-Fu Lee - one of
the world’s most respected experts on AI - reveals that China has suddenly
caught up to the US at an astonishingly rapid pace. As the US-Sino competition
begins to heat up, Lee envisions China and the US forming a powerful duopoly in
AI, but one that is based on each nation’s unique and traditional cultural
inclinations.
Building
upon his longstanding US-Sino technology career (working at Apple, Microsoft,
and Google) and his much-heralded New York Times Op-Ed from
June 2017, Dr. Lee predicts that Chinese and American AI will have a stunning
impact on not just traditional blue-collar industries but will also have a
devastating effect on white-collar professions. Is the concept of universal
basic income the solution? In Dr. Lee’s opinion, probably not.
In AI
Superpowers, he outlines how millions of suddenly displaced workers must
find new ways to make their lives meaningful, and how government policies will
have to deal with the unprecedented inequality between the "haves"
and the "have-nots." Even worse, Lee says the transformation to AI is
already happening all around us, whether we are aware of it or not.
Dr.
Lee - a native of China but educated in America - argues powerfully that these
unprecedented developments will happen much sooner than we think. He cautions
us about the truly dramatic upheaval that AI will unleash and how we need to
start thinking now on how to address these profound changes that are coming to
our world.
The Future of Artificial Intelligence and
its Impact on Society
Past Event —
November 3, 2017
Robots and
Artificial Intelligence: https://www.cfr.org/search?keyword=Robots%20and%20Artificial%20Intelligence
Intel’s
new ‘Loihi’ chip mimics neurons and synapses in the human brain
Automatically gets smarter over time
September 29, 2017
Intel announced this week a self-learning, energy-efficient neuromorphic
(brain-like) research chip codenamed “Loihi”* that mimics how the human brain
functions. Under development for six years, the chip uses 130,000 “neurons” and
130 million “synapses” and learns in real time, based on feedback from the
environment.**
Neuromorphic chip models are inspired by how neurons communicate and
learn, using spikes (brain pulses) and synapses capable of learning.
The idea is to help computers self-organize and make decisions based on
patterns and associations,” Michael Mayberry, PhD, corporate vice president and
managing director of Intel Labs at Intel Corporation, explained in a blog post.
He said the chip automatically gets smarter over time and doesn’t need to
be trained in the traditional way. He sees applications in areas that would
benefit from autonomous operation and continuous learning in an unstructured
environment, such as automotive, industrial, and personal-robotics areas.
For example, a cybersecurity system could identify a breach or a hack
based on an abnormality or difference in data streams. Or the chip could learn
a person’s heartbeat reading under various conditions — after jogging,
following a meal or before going to bed — to determine a “normal” heartbeat.
The system could then continuously monitor incoming heart data to flag patterns
that don’t match the “normal” pattern, and could be personalized for any user.
“Machine learning models such as deep learning have made tremendous
recent advancements by using extensive training datasets to recognize objects
and events. However, unless their training sets have specifically accounted for
a particular element, situation or circumstance, these machine learning systems
do not generalize well,” Mayberry notes.
The Loihi test chip
Loihi currently exists as a research test chip that offers flexible
on-chip learning and combines training and inference. Researchers have
demonstrated it learning at a rate that is a 1 million times improvement
compared with other typical spiking neural nets, as measured by total
operations to achieve a given accuracy when solving MNIST digit
recognition problems, Mayberry said. “Compared to technologies such as
convolutional neural networks and deep learning neural networks, the Loihi test
chip uses many fewer resources on the same task.”
Fabricated on Intel’s 14 nm process technology, the chip is also up to
1,000 times more energy-efficient than general-purpose computing required for
typical training systems, he added.
In the first half of 2018, IBM plans to share the Loihi test chip with
leading university and research institutions with a focus on advancing AI. The
goal is to develop and test several algorithms with high efficiency for
problems including path planning, constraint satisfaction, sparse coding,
dictionary learning, and dynamic pattern learning and adaptation.
“Looking to the future, Intel believes that neuromorphic computing offers
a way to provide exascale performance in a construct inspired by how the brain
works,” Mayberry said.
* “Loihi seamount, sometimes known as the ‘youngest
volcano’ in the Hawaiian chain, is an undersea mountain rising more than 3000
meters above the floor of the Pacific Ocean … submerged in the Pacific off of
the south-eastern coast of the Big Island of Hawaii.” — Hawaii
Center for Volcanology
** For comparison, IBM’s
TrueNorth neuromorphic chip currently has 1 million
neurons and 256 million synapses.
09.18.19
The company is aiming to adapt its
productivity suite to a short-attention-span world, bringing years of research
into products like PowerPoint, Outlook, and Excel…: https://www.fastcompany.com/90402486/how-human-curation-came-back-to-clean-up-ais-messes
Do our brains use the same kind of
deep-learning algorithms used in AI?
Bridging the gap
between neuroscience and AI
February 23, 2018
Deep-learning researchers
have found that certain neurons in the brain have shape and electrical
properties that appear to be well-suited for “deep learning” — the kind of
machine-intelligence used in beating humans at Go and Chess.
Canadian Institute For
Advanced Research (CIFAR) Fellow Blake Richards and his colleagues — Jordan Guerguiev at the University of Toronto, Scarborough, and Timothy Lillicrap at Google DeepMind — developed an algorithm
that simulates how a deep-learning network could work in our brains. It
represents a biologically realistic way by which real brains could do deep
learning.*
The finding is detailed in
a study published
December 5th in the open-access journal eLife. (The paper is highly
technical; Adam Shai of Stanford University and Matthew E. Larkum of Humboldt
University, Germany wrote a more accessible paper summarizing the ideas, published in the
same eLife issue.)
“Most of these neurons are shaped like trees,
with ‘roots’ deep in the brain and ‘branches’ close to the surface,” says
Richards. “What’s interesting is that these roots receive a different set of
inputs than the branches that are way up at the top of the tree.” That allows
these functions to have the required separation.
Using this knowledge of the
neurons’ structure, the researchers built a computer model using the same
shapes, with received signals in specific sections. It turns out that these
sections allowed simulated neurons in different layers to collaborate — achieving
deep learning.
“It’s just a set of
simulations so it can’t tell us exactly what our brains are doing, but it does
suggest enough to warrant further experimental examination if our own brains
may use the same sort of algorithms that they use in AI,” Richards says.
“No one has tested our
predictions yet,” he told KurzweilAI. “But, there’s a new preprint that
builds on what we were proposing in a nice way from Walter Senn‘s
group, and which includes some results on unsupervised learning (Yoshua [Bengio] mentions this work in his talk).
How the brain achieves deep learning
The tree-like pyramidal
neocortex neurons are only one of many types of cells in the brain. Richards
says future research should model different brain cells and examine how they
interact together to achieve deep learning. In the long term, he hopes
researchers can overcome major challenges, such as how to learn through
experience without receiving feedback or to solve the “credit assignment
problem.”**
Deep learning has brought
about machines that can “see” the world more like humans can, and recognize
language. But does the brain actually learn this way? The answer has the
potential to create more powerful artificial intelligence and unlock the
mysteries of human intelligence, he believes.
“What we might see in the
next decade or so is a real virtuous cycle of research between neuroscience and
AI, where neuroscience discoveries help us to develop new AI and AI can help us
interpret and understand our experimental data in neuroscience,” Richards says.
Perhaps this kind of
research could one day also address future ethical and other
human-machine-collaboration issues — including merger, as Elon Musk and Ray
Kurzweil have proposed, to achieve a “soft takeoff”
in the emergence of superintelligence.
* This research
idea goes back to AI pioneers Geoffrey Hinton, a CIFAR Distinguished Fellow and founder of the
Learning in Machines & Brains program, and program Co-Director Yoshua Bengio, who was one of the main motivations for founding
the program. These researchers sought not only to develop artificial
intelligence, but also to understand how the human brain learns, says Richards.
In the early
2000s, Richards and Lillicrap took a course with Hinton at the University of
Toronto and were convinced deep learning models were capturing “something real”
about how human brains work. At the time, there were several challenges to
testing that idea. Firstly, it wasn’t clear that deep learning could achieve
human-level skill. Secondly, the algorithms violated biological facts proven by
neuroscientists.
The paper builds
on research from Bengio’s lab on a more biologically plausible way
to train neural nets and
an algorithm developed by
Lillicrap that
further relaxes some of the rules for training neural nets. The paper also
incorporates research from Matthew Larkam on the structure of neurons in the
neocortex.
By combining
neurological insights with existing algorithms, Richards’ team was able to
create a better and more realistic algorithm for simulating learning in the
brain.
The study was
funded by the Natural Sciences and Engineering Research Council of Canada
(NSERC), a Google Faculty Research Award, and CIFAR.
** In the paper,
the authors note that a large gap exists between deep learning in AI and our
current understanding of learning and memory in neuroscience. “In particular,
unlike deep learning researchers, neuroscientists do not yet have a solution to
the ‘credit assignment problem’ (Rumelhart et al., 1986; Lillicrap et al., 2016; Bengio et al., 2015).
Learning to optimize some behavioral or cognitive function requires a method
for assigning ‘credit’ (or ‘blame’) to neurons for their contribution to the
final behavioral output (LeCun et al., 2015; Bengio et al., 2015).
The credit assignment problem refers to the fact that assigning credit in
multi-layer networks is difficult, since the behavioral impact of neurons in
early layers of a network depends on the downstream synaptic connections.” The
authors go on to suggest a solution.: http://www.kurzweilai.net/do-our-brains-use-the-same-kind-of-deep-learning-algorithms-used-in-ai?utm_source=KurzweilAI+Weekly+Newsletter&utm_campaign=8a3132f577-UA-946742-1&utm_medium=email&utm_term=0_147a5a48c1-8a3132f577-282212701
Gloucester Daily Times February 8, 2019
author: Anthony
J. Marolda
—
introduction —
Many
people are concerned about climate change and its impact on the planet over the
next 80 years. However, another threat to humanity that is much more alarming,
very real + very close — is the creation of artificial super-intelligence (the
abbreviation is ASI). Today’s artificial intelligence will become ASI when it’s
billions of times “smarter” than people. We’ll need to learn to live with ASI,
or we might not be around much longer.
—
technology execs say concerns are real —
Many
high-level executives in the tech industry warn about the dangers of ASI — for
example: Sundar Pichai, CEO at Google. Google is a leader in the development +
application of computer software in the field of artificial intelligence. For
example: the company Deep Mind — under the Alphabet co. umbrella that also owns
Google — is developing software programs that learn to solve complex problems,
without teaching them how. This is the true beginning of ASI: it will have
characteristics of human intelligence but will astronomically exceed it.
Sundar
Pichai gave an interview to the Washington Post — he said
artificial intelligence holds great promise to benefit humanity. But some
scientists worry about potential harmful applications of the tech. Pichai said
their concerns are “very legitimate.” For example, he described autonomous AI
weapons that can make “kill decisions” on their own. Think about the fictional
SkyNet system from the Terminator series of films.
Other
tech executives agree with Pichai about the serious possible threat from
artificial intelligence. Elon Musk — inventor of electric cars by Tesla co. and
rockets by Space X co. — spoke in the documentary film: Do You Trust This Computer? He said ASI can
leave humanity behind, leading to the creation of an “immortal dictator” who’ll
control the world.
Elon
Musk said: “At least when there’s an evil (human) dictator, that person is
going to die. But for an ASI (software) dictator, there’d be no death. It would
live forever. And then we’d have an immortal dictator we can never escape. If
ASI has a goal and humanity just happens to be in the way, it will destroy
humanity as a matter of course — without even thinking about it. No hard
feelings. Like if we’re building a road, and an anthill happens to be in the
way: we don’t hate ants. We’re building a road, and so goodbye anthill.”
—
DeepMind by Alphabet co. develops powerful software —
Google’s
company Deep Mind has achieved a turning point in creating human-like
artificial intelligence. They have a computer software program called
AlphaZero that shows human-like qualities of intuition + creativity. AlphaZero
was made with the ability to learn + remember what it does.
For
example, the developers set-up the program to learn to play chess. Unlike past
computers that were programmed by the developers to play the game — AlphaZero
knew nothing about chess, except the basic rules. To learn, it played 44
million matches with itself in 9 hours and learned from each one. Eventually it
was able to beat chess grand-masters, but with approaches never before used by
a chess computer. It was exhibiting human-like intuition and creativity.
The
former world chess champion Garry Kasparov said: “Instead of processing
human instructions and knowledge at tremendous speed — the way all other
previous chess machines did — AlphaZero generates its own knowledge. It plays
with a dynamic style, similar to mine. The implications go beyond the chess
board.”
AlphaZero
by DeepMind is a sign that we’re approaching a theoretical time in history
known as the “technological singularity.” Experts say that’s the point when:
“the invention of artificial super-intelligence will suddenly trigger run-away
technological growth, resulting in unfathomable changes to human civilization.”
That’s is also the point when humans could lose control of society, with no
chance of getting it back.
—
Ray Kurzweil + Elon Musk say humans must merge with computers —
Ray
Kurzweil is a famous inventor, futurist, best selling author, and entrepreneur.
He’s currently a director of engineering at Google. He’s done a lot of thinking
about the coming technological singularity. His track record making predictions
about the future is about 86 percent correct. And, based on the exponential
rate of progress in the AI industry, Kurzweil calculates the singularity will
happen in year 2029. He anticipates by that time, ASI systems will be “billions
of times smarter” than humans.
Elon
Musk and Ray Kurzweil agree. Humans must somehow merge with computers — to stay
relevant in the world of ASI. Musk has formed a company called Neuralink to
achieve that goal. Neuralink is developing ultra-high bandwidth (the speed +
capacity of the connection), implantable, brain-machine interfaces to connect
humans with computers. Musk is hoping to demonstrate the tech by year 2019.
Kurzweil
has a similar vision to resolve the threat posed by ASI. He foresees a
computer-mind connection, but a different type than Musk’s concept. He calls it
a neo-cortex connection, made using nano-bots — molecule sized devices injected
into the bloodstream to accomplish pre-programmed tasks. Kurzweil’s idea is to
use the nano-bots to connect your brain directly to the web, upgrading your
intelligence and memory capacity by orders of magnitude. Thus, as the machines
become smarter, so do humans.
—
humans merging with machines —
But
how far along is the development of nano-sized robotic systems that can enter
the human body? The first actual use of a miniature drug delivery system was
tested by researchers at Arizona State University. They created cell-sized bots
— made of sheets of biological molecules (not machines) — and injected them
into the bloodstream of mice with cancer. The bots went directly to the
cancer’s tumor cells and injected them with blood clotting drugs to cut off
their blood supply and stop the tumor’s growth. It functioned, shrinking the
tumors. Over the next 10 years, nano-bot tech could grow at a fast rate —
getting closer to Kurzweil’s vision of connecting the human neo-cortex to the
cloud.
So
instead of humans becoming obsolete, we could be working with the machines. But
it’s important that humanity perfect the human-machine connection tech before
the singularity — the point where AI becomes ASI. If we cross that horizon and
we’re not working symbiotically with our machines, we may not be given the
chance later.
3 ноября 2017
The
Future of Artificial Intelligence and its Impact on Society
Ray Kurzweil
keynote presentation at Council on Foreign Relations meeting
November 3, 2017
June 8, 2017 by Nina Larson
Sophia, a humanoid robot, is the main attraction at a conference on
artificial intelligence this week but her technology has raised concerns for
future human jobs
Sophia smiles mischievously, bats her eyelids and tells a joke. Without the
mess of cables that make up the back of her head, you could almost mistake her
for a human.
The humanoid robot,
created by Hanson robotics, is the main attraction at a UN-hosted conference in
Geneva this week on how artificial intelligence can
be used to benefit humanity.
The event comes as concerns grow that rapid advances in such technologies
could spin out of human control and become detrimental to society.
Sophia herself insisted "the pros outweigh the cons" when it
comes to artificial intelligence.
"AI is good for the world, helping people in various ways," she
told AFP, tilting her head and furrowing her brow convincingly.
Work is underway to make artificial intelligence "emotionally smart,
to care about people," she said, insisting that "we will never
replace people, but we can be your friends and helpers."
But she acknowledged that "people should question the consequences of new technology."
Among the feared consequences of the rise of the robots is the growing
impact they will have on human jobs and economies.
Legitimate concerns
Decades of automation and robotisation have already revolutionised the industrial
sector, raising productivity but cutting some jobs.
AI developers
hope their inventions will some day revolutionise sectors such as healthcare
and education, especuially in rural areas with shortages of professional
And now automation and AI are expanding rapidly into other sectors, with
studies indicating that up to 85 percent of jobs in developing countries could
be at risk.
"There are legitimate concerns about the future of jobs, about the
future of the economy, because when businesses apply automation, it tends to
accumulate resources in the hands of very few," acknowledged Sophia's
creator, David Hanson.
But like his progeny, he insisted that "unintended consequences, or
possible negative uses (of AI) seem to be very small compared to the benefit of
the technology."
AI is for instance expected to revolutionise healthcare and education,
especially in rural areas with shortages of doctors and teachers.
"Elders will have more company, autistic children will have endlessly
patient teachers," Sophia said.
But advances in robotic technology have sparked growing fears that humans
could lose control.
Killer robots
Amnesty International chief Salil Shetty was at the conference to call for
a clear ethical framework to ensure the technology is used on for good.
"We need to have the principles in place, we need to have the checks
and balances," he told AFP, warning that AI is "a black box... There
are algorithms being written which nobody understands."
AI
technology is already being used in the US for 'predictive policing' and rights
groups are calling for regulation
Shetty voiced particular concern about military use of AI in weapons and
so-called "killer robots".
"In theory, these things are controlled by human beings, but we don't
believe that there is actually meaningful, effective control," he said.
The technology is also increasingly being used in the United States for
"predictive policing", where algorithms based on historic trends
could "reinforce existing biases" against people of certain
ethnicities, Shetty warned.
Hanson agreed that clear guidelines were needed, saying it was important to
discuss these issues "before the technology has definitively and
unambiguously awakened."
While Sophia has some impressive capabilities, she does not yet have
consciousness, but Hanson said he expected that fully sentient machines could
emerge within a few years.
"What happens when (Sophia fully) wakes up or some other machine,
servers running missile defence or managing the stock market?" he asked.
The solution, he said, is "to make the machines care about us."
"We need to teach them love."
Read more at: https://phys.org/news/2017-06-ai-good-world-ultra-lifelike-robot.html#jCp
deck: Our singular future.
deck: The next 20 years will change our idea of what it is to be human.
author: by Robert Levine date: July 2006
The New Human
deck: Our singular future.
deck: The next 20 years will change our idea of what it is to be human.
author: by Robert Levine date: July 2006
In the mid-1980s inventor
Ray Kurzweil predicted that a few interconnected computers used by scientists
would serve as the basis for a worldwide communications network. AI the time it
seemed far-fetched, but Arpanet evolved into the Internet. Kurzweil
subsequently postulated the law of accelerating returns, which holds that
information technology increases exponentially, doubling every year.
He later predicted that
computers would exceed human intelligence, eventually reaching a point-the
singularity-at which civilization would fundamentally be transformed. In his
new book, The Singularity Is Near, Kurzweil explores
the implications of that change. He believes our bodies will evolve as much as
our machines. In fact, he predicts a clear separation will no longer exist
between the two.
Ray Kurzweil says: “If you
describe what human beings enhanced with this technology will be capable of
some decades in from now — they’d appear like gods to us today.”
1. What is the
singularity?
It’s a metaphor borrowed
from physics, which in turn had borrowed it from mathematics. In physics it’s a
point of profound transformation, a rupture in space-time. There’s an event
horizon around It that’s hard to see into. But the historical singularity is an
event that will occur, in my estimation, in about 49 years. It will be a
profound transformation of human civilization caused by the emergence of
non-biological intelligence billions of times more powerful than un-enhanced
biological intelligence.
Underlying all this is the
observation that information technology grows exponentially. Bandwidth, the
price-performance ratio of computers and the size of the Internet all double
every year. That’s true of all kinds of information. For example, the amount of
DNA sequencing we’re doing doubles every year. The resolution of brain scanning
doubles every year.
2. What if there’s a
limit to Moore’s law, which says computing power doubles every 18 months?
Certain things follow
predictable trends. Moore’s law will reach a limit, it’s estimated, in 2020.
But every time we’ve come to the end of one paradigm, we reach another. Moore’s
law was the fifth paradigm. The third was vacuum tubes; the fourth was
transistors. The sixth wilt be three-dimensional molecular circuits. One cubic
Inch of nanotube circuitry, for example. would be 100 million times more
powerful than the human brain.
3. But will it be
better than the human brain?
We’ll take the power of
human intelligence-our ability to recognize patterns—and combine it with the
ways machines are already superior. They can remember billions of things and
share information at high speeds. So non-human intelligence will ultimately be
able to rea all human literature and learn all science and technology. Plus it
will examine its own source code and redesign itself, which is something we as
humans aren’t able to do.
4. In your book The
Singularity Is Near you write, “Our technology will match and then exceed the
refinement and suppleness of what we regard as the best of human traits.” So
you’re not talking about just calculations per second?
Our emotional intelligence
is the cutting edge of human intelligence. Humor and love are complicated
behaviors. Understanding them requires a very high level of intelligence.
5. Are you saying
love can be reduced to calculations? If you’re right, how will that change the
way we look at what it means to be human?
That’s precisely why this is
called the singularity — because it’s so hard to wrap our mind around. We take
for granted certain characteristics of software that are actually advantages we
don’t have as human beings. If you change computers, you don’t throw all your
files away; you just port them over to the new hardware.
The information has a
longevity that transcends the hardware it’s on. But that’s not the case with
another important file, the mind. We take for granted that when our hardware
crashes, the software is gone with it. There’s no reason to imagine the mind
can’t transcend the hardware it runs on.
6. Is that why you
way that in 25 years we’ll be more non-biological than biological?
Computers used to be remote;
now they’re in our pockets. They’ll make their way into our clothing. They’ll
make heir way into our body and brain. You can’t point to a single organ for
which we haven’t made enhancements or started work on them. Some Parkinson’s
dissease patients have an FDA approved neural implant.
The latest generation of it
allows you to upload software from outside the patient. If we follow this trend
— exponential shrinking of technology — we’ll be able to send intelligent
nanobots the size of blood cells into our brain. Neural implants introduced
non-invasively will be able to extend our intelligence or provide virtual
reality by replacing the input from our senses so it feels as if we’re in a
different environment.
7. What if people
don’t want to become more non-biological? What I they can’t afford it?
There are always early and
late adopters, but I think it’s going to be a slippery slope. Some conservative
applications will be about just keeping you healthy and doing routine things
like expanding your memory. Very few people will eschew those enhancements.
How many people won’t use
eyeglasses? When technology is introduced , only the wealthy can afford it and
it doesn’t work well. A few years go by, and it’s expensive and works a bit
better. Eventually it’s not that expensive and works well. Not so long ago, if someone
took out a mobile phone, it meant he was a member of the power elite.
8. And you think all
this technology will radically extend human life?
In the book I wrote with
Terry Grossman MD — Fantastic Voyage: Live Long Enough to Live Forever —
we talk about bridges to extreme life extension. Bridge one is what we can do
today. I think people from 40 to 80, maybe a little older, can extend their
longevity by hanging in there for a little longer. The point of bridge one is
to be in good shape for 10 or 15 years, when bridge two comes along.
9. How do you do
that?
Aging is not one thing; it’s
a number of processes. We have strategies for slowing down each of the dozen
aging processes. The program we prescribe depends on which health issues you
have. Disease doesn’t come out of the blue. You can catch it. Find out where
you are in certain measurements of health before you get cancer or a heart
attack, a third of which are fatal.
10. What happens if
you make it to bridge 2?
Bridge two will be the
mastery of our biology, being able to turn genes on and off. One of those
genes, the fat-insulin receptor gene, says, “Hold on to every calorie because
the next hunting season may not turn out so well.” We’d like to turn that off.
That technology will reach maturity in 10 to 15 years. this will bring us to
the third bridge, which is nano-tech, with which we can not just refine and
reprogram biology but go beyond it altogether.
One super app is nanobots,
blood-cells-size devices that can go inside our body and brain and keep us
healthy. We have already put microscopic machines into animals. If you apply
these exponential trends, which I maintain are quite predictable, we’ll be able
to have sophisticated computerized devices in our bloodstream, performing very
sophisticated functions.
11. But the notion
that life is limited has always been one of the principles that define what it
means to be human.
I don’t think we need death
to give life meaning. There are different concepts of what it means to be
human. My concept is different: We’re a species that goes beyond our
limitations. We didn’t stay on the ground, we didn’t stay on the planet, and we
didn’t stay within the limitations of our biology. Extending human longevity is
not a new story. Human life expectancy was 37 in 1800. Sanitation and
antibiotics brought it into the 60’s and now it’s in the 80s. We’ll have
another major jump in longevity when we reprogram our genes, turning off genes
with RNA interference, turning on genes with gene therapy, turning enzymes on
and off — things I believe we’ll master in 15 years.
12. Will this make
us happier?
I’m not confident we will
overcome human conflict. Some people think that because I talk about this
technology’s problem-solving ability, it is a utopian vision. But I think we
will introduce new problems along the way. Also, I don’t think that just being
happy is the right goal.
A salamander may happy, but
its life is not very interesting compared with our life. Would you rather be a
happy salamander or have a dynamic life of accomplishment and challenge? The
meaningful thing in life is creating knowledge. I don’t just mean random bits
of data but knowledge — art, music, poetry, literature or even our
relationships and the way we express ourselves.
13. What will human
sexuality be like in 20 or 25 years?
These technologies will have
a profound impact because sex and intimacy involve all five senses. By 2020
we’ll have perfected virtual reality that can be delivered from outside the
body. We’ll have images written to our retinas, and we’ll be able to enter a
full-immersion virtual reality environment. So you could be with someone else
from a sensory perspective. You’ll feel as though you’re really with tat
person. You could take a walk on a virtual beach.
The whole idea of what it
means to have a sexual + romantic relationship will be different. But what’s
really interesting is that we’ll eventually have virtual reality from inside
the nervous system. We’ll have nano-bots that go inside the brain, block-out
signals coming from your senses and replace them with the senses your brain
would be receiving if you were in a virtual environment. You could go to this
environment with one other person and have a romantic encounter involving all 5
senses. You could be someone else. A couple could turn themselves into each
other. Ultimately it will be highly realistic — and competitive with reality.
special section |
Future Shock
The legendary physicist
Niels Bohr PhD said: “Prediction is very difficult, especially of the future.”
But according to futurists, we’re on the verge of astonishing developments.
Here are 4 innovations we should see in the next decade.
1. | enhanced eyes
Some time after 2010 active
contact lenses will be used to produce computer-generated overlays on what we
see in the real world. “Even if your partner’s physical appearance is not quite
up to your hopes,” writes Ian Pearson, futurist for British Telecom group, “it
could be digitally enhanced with something closer to your dreams.”
2. | my robot
By 2010 all-purpose robots
should be available for common household tasks. Automobile manufacturers such
as Honda + Toyota will lead the way. Hans Moravec of Carnegie Mellon predicts
that by 2025 the robot market will be larger than the market for automobiles.
3. | digitized physical
objects
The molecular patterns of
everyday objects will be scanned and coded as digital information. Once the
make-up and structure of an object exists as a file, we can print-out complex
objects composed of many types of organic + non-organic materials to create new
formations or modules that can include embedded capabilities like
bio-electronics, bio-processes, or compartmentalized functions.
Futurist Jeff Harrow said:
“An example would be 3D printing of organic tissue. The day will come when you
can replicate — on a printer — a new donor liver that won’t be rejected by the
patient.”
4. | enhanced skin
According to Pearson, by the
end of this decade we will be able to build ID and memory chips, sensors and
short-range communications devices smaller than human skin cells. These will be
printed on or blasted into the upper layers of the skin and arranged into
circuits so that electronic devices such as cell phones, keyboards and MP3
players can be embedded into your forearm , the back of your hand or your wrist.
Being Human in the Age of Artificial Intelligence
August 2, 2017
author Max Tegmark
year published 2017
The robot takeover will ignite an explosion of
“awe-inspiring” life even if humans don’t survive, according to this
exhilarating, demoralizing primer. MIT physicist Tegmark (Our Mathematical
Universe) surveys advances in artificial intelligence such as self-driving
cars and Jeopardy-winning software, but focuses on the looming
prospect of “recursive self-improvement”—AI systems that build smarter versions
of themselves at an accelerating pace until their intellects surpass ours.
Tegmark’s smart, freewheeling discussion leads to fascinating speculations on
AI-based civilizations spanning galaxies and eons—and knotty questions: Will
our digital overlords be conscious? Will they coddle us with abundance and
virtual-reality idylls or exterminate us with bumblebee-size attack robots?
While digerati may be enthralled by the idea of superintelligent civilizations
where “beautiful theorems” serve as the main economic resource, Tegmark’s
future will strike many as a one in which, at best, humans are dependent on
AI-powered technology and, at worst, are extinct. His call for strong controls
on AI systems sits awkwardly beside his acknowledgment that controlling such
godlike entities will be almost impossible. Love it or hate it, it’s an
engrossing forecast.
Campaign to Stop Killer Robots | Slaughterbots
|
OpenAI’s GPT-4 is so powerful that experts want to slam the
brakes on generative AI |
|
We
can keep developing more and more powerful AI models, but should we? Experts
aren’t so sure OpenAI
GPT-4 ir tik jaudīgs, ka eksperti vēlas nospiest ģeneratīvo AI Mēs
varam turpināt izstrādāt arvien jaudīgākus AI modeļus, bet vai mums
vajadzētu? Eksperti nav tik pārliecināti fastcompany.com/90873194/chatgpt-4-power-scientists-warn-pause-development-generative-ai-letter |
Will artificial intelligence become
conscious?
December 22, 2017
Forget about today’s modest incremental advances in artificial intelligence, such as the
increasing abilities of cars to drive themselves. Waiting in the wings might be a groundbreaking development: a machine that is aware of itself and its
surroundings, and that could take in and process massive amounts of data in
real time. It could be sent on dangerous missions, into space or combat. In
addition to driving people around, it might be able to cook, clean, do laundry
— and even keep humans company when other people aren’t nearby.
A particularly advanced set
of machines could replace humans at literally all jobs. That would save
humanity from workaday drudgery, but it would also shake many societal
foundations. A life of no
work and only play may turn out to be a dystopia.
Conscious machines would
also raise troubling legal and ethical
problems. Would a conscious machine
be a “person” under law and be liable if its actions hurt someone, or if something
goes wrong? To think of a more frightening scenario, might these machines rebel
against humans and wish to eliminate us altogether? If yes, they represent the culmination of
evolution.
As a professor of electrical
engineering and computer science who works in machine learning and quantum
theory, I can say that researchers are divided on whether these sorts of
hyperaware machines will ever exist. There’s also debate about whether machines
could or should be called “conscious” in the way we think of humans, and even
some animals, as conscious. Some of the questions have to do with technology;
others have to do with what consciousness actually is.
Is awareness enough?
Most computer scientists
think that consciousness is a characteristic that will emerge as technology
develops. Some believe that consciousness involves accepting new information,
storing and retrieving old information and cognitive processing of it all
into perceptions and actions. If that’s right, then one day machines will indeed
be the ultimate consciousness. They’ll be able to gather more information than a
human, store more than many libraries, access vast databases in milliseconds
and compute all of it into decisions more complex, and yet more logical, than any person ever could.
On the other hand, there are
physicists and philosophers who say there’s something more about human behavior
that cannot be computed by a machine. Creativity, for example, and the sense of
freedom people possess don’t appear to come from logic or calculations.
Yet these are not the only
views of what consciousness is, or whether machines could ever achieve it.
Quantum views
Another viewpoint on
consciousness comes from quantum theory, which is the deepest theory of
physics. According to the orthodox Copenhagen Interpretation, consciousness and the physical world are
complementary aspects of the same reality. When a person observes, or
experiments on, some aspect of the physical world, that person’s conscious interaction causes
discernible change. Since it takes
consciousness as a given and no attempt is made to derive it from physics, the Copenhagen Interpretation may be called the
“big-C” view of consciousness, where it is a thing that exists by itself –
although it requires brains to become real. This view was popular with the
pioneers of quantum theory such as Niels Bohr, Werner Heisenberg and Erwin
Schrödinger.
The interaction between
consciousness and matter leads to paradoxes that remain unresolved after 80
years of debate. A well-known example of this is the paradox of Schrödinger’s cat, in which a cat is placed in a situation that
results in it being equally likely to survive or die – and the act of
observation itself is what makes the outcome
certain.
The opposing view is that
consciousness emerges from biology, just as biology itself emerges from
chemistry which, in turn, emerges from physics. We call this less expansive concept of
consciousness “little-C.” It agrees with the neuroscientists’ view that
the processes of the mind are
identical to states and processes of the brain. It also agrees with a more recent interpretation
of quantum theory motivated by an attempt to rid it of paradoxes, the Many Worlds Interpretation, in which observers are a part of the mathematics
of physics.
Philosophers of science
believe that these modern quantum physics views of consciousness have parallels
in ancient philosophy. Big-C is like the theory of mind in Vedanta – in which
consciousness is the fundamental basis of reality, on par with the physical universe.
Little-C, in contrast, is
quite similar to Buddhism. Although the Buddha chose not to address the
question of the nature of consciousness, his followers declared that mind and
consciousness arise out of emptiness or nothingness.
Big-C and scientific discovery
Scientists are also
exploring whether consciousness is always a computational process. Some
scholars have argued that the creative moment is not at the end of a deliberate
computation. For instance, dreams or visions are supposed to have inspired Elias Howe‘s 1845 design of the modern sewing machine, and
August Kekulé’s discovery of the structure of benzene in 1862.
A dramatic piece of evidence
in favor of big-C consciousness existing all on its own is the life of
self-taught Indian mathematician Srinivasa Ramanujan, who died in 1920 at the age of 32. His notebook,
which was lost and forgotten for about 50 years and published only in 1988, contains several thousand formulas, without proof
in different areas of mathematics, that were well ahead of their time.
Furthermore, the methods by which he found the formulas remain elusive. He himself claimed that they were revealed to him
by a goddess while he was asleep.
The concept of big-C
consciousness raises the questions of how it is related to matter, and how
matter and mind mutually influence each other. Consciousness alone cannot make
physical changes to the world, but perhaps it can change the probabilities in
the evolution of quantum processes. The act of observation can freeze and even influence atoms’ movements, as
Cornell physicists proved in 2015. This may very well be an explanation of how matter
and mind interact.
Mind and self-organizing systems
It is possible that the
phenomenon of consciousness requires a self-organizing system, like the brain’s
physical structure. If so, then current machines will come up short.
Scholars don’t know if
adaptive self-organizing machines can be designed to be as sophisticated as the
human brain; we lack a mathematical theory of computation for systems like
that. Perhaps it’s true that only biological machines can be sufficiently
creative and flexible. But then that suggests people should – or soon will –
start working on engineering new biological structures that are, or could
become, conscious.: http://www.kurzweilai.net/will-artificial-intelligence-become-conscious?utm_source=KurzweilAI+Weekly+Newsletter&utm_campaign=e7c90f0a20-UA-946742-1&utm_medium=email&utm_term=0_147a5a48c1-e7c90f0a20-282212701
- 09.20.19
The rise of AI has led to tattered privacy
protections and rogue algorithms. Here’s what we can do about it.
This
article is part of Fast Company’s editorial series The New Rules of AI.
More than 60 years into the era of artificial intelligence, the world’s largest
technology companies are just beginning to crack open what’s possible with
AI—and grapple with how it might change our future. Click here to read all the stories in the
series.
Consumers and activists are rebelling against
Silicon Valley titans, and all levels of government are probing how they
operate. Much of the concern is over vast quantities of data that tech
companies gather—with and without our consent—to fuel artificial intelligence
models that increasingly shape what we see and influence how we act.
If “data is the new oil,” as boosters of the AI
industry like to say, then scandal-challenged data companies like Amazon, Facebook, and Google may face the same mistrust as oil
companies like BP and Chevron. Vast computing facilities refine crude data into
valuable distillates like targeted advertising and product recommendations. But
burning data pollutes as well, with faulty algorithms that make judgments on
who can get a loan, who gets hired and fired, even who goes to jail.
The extraction of crude data can be equally
devastating, with poor communities paying a high price. Sociologist and
researcher Mutale Nkonde fears that the poor will sell for cheap the rights to
biometric data, like scans of their faces and bodies, to feed algorithms for
identifying and surveilling people. “The capturing and encoding of our
biometric data is going to probably be the new frontier in creating value for
companies in terms of AI,” she says.
The further expansion of AI is inevitable, and it
could be used for good, like helping take violent images off the internet or speeding up the drug discovery process. The question
is whether we can steer its growth to realize its potential benefits while
guarding against its potential harms. Activists will have different notions of
how to achieve that than politicians or heads of industry do. But we’ve sought
to cut across these divides, distilling the best ideas from elected officials,
business experts, academics, and activists into five principles for tackling
the challenges AI poses to society.
1.
CREATE AN FDA FOR ALGORITHMS
Algorithms are impacting our world in powerful but
not easily discernable ways. Robotic systems aren’t yet replacing soldiers as
in The Terminator, but instead they’re slowly supplanting the
accountants, bureaucrats, lawyers, and judges who decide benefits, rewards, and
punishment. Despite the grown-up jobs AI is taking on, algorithms continue to
use childish logic drawn from biased or incomplete data.
Cautionary tales abound, such as a seminal 2016 ProPublica investigation that
found law enforcement software was overestimating the chance that black
defendants would re-offend, leading to harsher sentences. In August, the ACLU
of Northern California tested Rekognition, Amazon’s facial-recognition
software, on images of California legislators. It matched 26 of 120 state lawmakers to images
from a set of 25,000 public arrest photos, echoing a test the ACLU did of
national legislators last year. (Amazon disputes the ACLU’s methodology.)
Faulty algorithms charged with major
responsibilities like these pose the greatest threat to society—and need the
greatest oversight. “I advocate having an FDA-type board where, before an
algorithm is even released into usage, tests have been run to look at impact,”
says Nkonde, a fellow at Harvard University’s Berkman Klein Center for Internet
& Society. “If the impact is in violation of existing laws, whether it be
civil rights, human rights, or voting rights, then that algorithm cannot be
released.”
Nkonde is putting that idea into practice by
helping write the Algorithmic Accountability Act of 2019, a bill
introduced by U.S. Representative Yvette Clarke and Senators Ron Wyden and Cory
Booker, all of whom are Democrats. It would require companies that use AI to
conduct “automated decision system impact assessments and data protection
impact assessments” to look for issues of “accuracy, fairness, bias,
discrimination, privacy, and security.”
These would need to be in plain language, not
techno-babble. “Artificial intelligence is . . . a very simple concept, but
people often explain it in very convoluted ways,” says Representative Ro
Khanna, whose Congressional district contains much of Silicon Valley. Khanna
has signed on to support the Algorithmic Accountability Act and is a co-sponsor of a resolution calling for national
guidelines on ethical AI development.
Chances are slim that any of this legislation will
pass in a divided government during an election year, but it will likely
influence the discussion in the future (for instance, Khanna co-chairs Bernie
Sanders’s presidential campaign).
2.
OPEN UP THE BLACK BOX OF AI FOR ALL TO SEE
Plain-language explanations aren’t just wishful
thinking by politicians who don’t understand AI, according to someone who
certainly does: data scientist and human rights activist Jack Poulson.
“Qualitatively speaking, you don’t need deep domain expertise to understand
many of these issues,” says Poulson, who resigned his position at Google to protest its
development of a censored, snooping search engine for the Chinese market.
To understand how AI systems work, he says, civil
society needs access to the whole system—the raw training data, the algorithms
that analyze it, and the decision-making models that emerge. “I think it’s
highly misleading if someone were to claim that laymen cannot get insight from
trained models,” says Poulson. The ACLU’s Amazon Rekognition tests, he says,
show how even non-experts can evaluate how well a model is working.
AI can even help evaluate its own failings, says
Ruchir Puri, IBM Fellow and the chief scientist of IBM Research who
oversaw IBM’s AI platform Watson from 2016 to 2019. Puri has an intimate
understanding of AI’s limitations: Watson Health AI came under fire from
healthcare clients in 2017 for not delivering the intelligent diagnostic help promised—at
least not on IBM’s optimistic timeframe.
“We are continuously learning and evolving our
products, taking feedback, both from successful and, you know,
not-so-successful projects,” Puri says.
IBM is trying to bolster its reputation as a
trustworthy source of AI technology by releasing tools to help make it easier
to understand. In August, the company released open-source software to analyze and explain how algorithms come to their decisions.
That follows on its open-source software from 2018 that looks for bias in data used to train AI models,
such as those assigning credit scores.
“This is not just, ‘Can I explain this to a data
scientist?'” says Puri. “This is, ‘Can I explain this to someone who owns a
business?'”
3.
VALUE HUMAN WISDOM OVER AI WIZARDRY
The overpromise of IBM Watson indicates another
truth: AI still has a long way to go. And as a result, humans should remain an
integral part of any algorithmic system. “It is important to have humans in the
loop,” says Puri.
Part of the problem is that artificial intelligence
still isn’t very intelligent, says Michael Sellitto, deputy director of
Stanford University’s Institute for Human-Centered
Artificial Intelligence (HAI). “If you take an algorithm out of
the specific context for which it was trained, it fails quite spectacularly,”
he says.
That’s also the case when algorithms are poorly
trained with biased or incomplete data—or data that doesn’t prepare them for
nuance. Khanna points to Twitter freezing the account of Senate Majority Leader Mitch
McConnell’s campaign for posting a video of people making
“violent threats.” But they were protestors against McConnell, whose team was
condemning the violent threats, not endorsing them.
Because of AI’s failings, human judgment will
always have to be the ultimate authority, says Khanna. In the case of Twitter’s
decision to freeze McConnell’s account, “it turns out that the context
mattered,” he says. (It’s not clear if Twitter’s decision was based on
algorithms, human judgment, or both.)
But the context of humans making decisions also
matters. For instance, Khanna is collaborating with Stanford HAI to develop a
national AI policy framework, which raises its own questions of bias. The
economy of Khanna’s district depends on the AI titans, whose current and former
leaders dominate HAI’s Advisory Council. Industry leaders who have bet their
future on AI will likely have a hard time making fair decisions that benefit
everyone, not just businesses.
“That’s why I am putting so much effort into
advocating for them to have more members of civil society in the room and for
there to be at least some accountability,” says Poulson. He led a petition against an address by former Google
CEO Eric Schmidt that has been planned for HAI’s first major conference in October.
Stanford has since added two speakers—Algorithmic
Justice League founder Joy Buolamwini and Stony Brook University art
professor Stephanie Dinkins—whom Poulson considers to be “unconflicted.”
(Stanford says that it was already recruiting the two as speakers before
Poulson’s petition.)
Humans are also making their voices heard within
big tech companies as well. Poulson is one of many current and former Googlers
to sound the alarm about ethical implications of the company’s tech
development, such as the Maven program to provide AI to the Pentagon. And tech worker activism is on the rise at other
big AI powerhouses, such as Amazon and Microsoft.
4.
MAKE PRIVACY THE DEFAULT
At the heart of many of these issues is privacy—a
value that has long been lacking in Silicon Valley. Facebook founder Mark
Zuckerberg’s motto, “Move fast and break things,” has been the modus operandi
of artificial intelligence, embodied in Facebook’s own liberal collection of
customer data. Part of the $5 billion FTC settlement against Facebook was for
not clearly informing users that it was using facial-recognition technology on their
uploaded photos. The default is now to exclude users from face scanning unless
they choose to participate. Such opt-ins should be routine across the tech
industry.
“We need a regulatory framework for data where,
even if you’re a big company that has a lot of data, there are very clear
guidelines about how you can use that data,” says Khanna.
That would be a radical shift for Big Tech’s
freewheeling development of AI, says Poulson, especially since companies tend to
incentivize quick-moving development. “The way promotions work is based upon
products getting out the door,” he says. “If you convince engineers not to
raise complaints when there is some fundamental privacy or ethics violation,
you’ve built an entire subset of the company where career development now
depends upon that abuse.”
In an ideal world, privacy should extend to never
collecting some data in the first place, especially without consent. Nkonde
worked with Representative Yvette Clarke on another AI bill, one that would prohibit the use of
biometric technology like face recognition in public housing. Bernie Sanders has
called for a ban on facial recognition in policing. California is
poised to pass a law that bans running facial recognition
programs on police body camera footage. San Francisco, Oakland, and Somerville,
Massachusetts, have banned facial recognition technology by city
government, and more cities are likely to institute their own bans. (Still,
these are exceptions to widespread use of facial recognition by cities
across the United States.)
Tech companies tend to argue that if data is
anonymized, they should have free reign to use it as they see fit.
Anonymization is central to Khanna’s strategy to compete with China’s vast data
resources.
But it’s easy to recover personal information from
purportedly anonymized records. For instance, a Harvard study found that 87% of
Americans can be identified by their unique combination of birth date, gender,
and zip code. In 2018, MIT researchers identified Singapore residents by analyzing overlaps in anonymized data sets of
transit trips and mobile phone logs.
5.
COMPETE BY PROMOTING, NOT INFRINGING, CIVIL RIGHTS
The privacy debate is central to the battle between
tech superpowers China and the United States. The common but simplistic view of
machine learning is that the more data, the more accurate the algorithm.
China’s growing AI prowess benefits from vast, unfettered information
collection on 1.4 billion residents, calling into doubt whether a country with
stricter privacy safeguards can amass sufficient data to compete.
But China’s advantage comes at a huge price,
including gross human rights abuses, such as the deep surveillance of the Uighur Muslim minority.
Omnipresent cameras tied to facial recognition software help track residents,
for instance, and analysis of their social relationships are used to assess
their risk to the state.
Chinese citizens voluntarily give up privacy far
more freely than Americans do, according to Taiwanese-American AI expert and
entrepreneur Kai-Fu Lee, who leads the China-based VC firm Sinovation Ventures.
“People in China are more accepting of having their faces, voices, and shopping
choices captured and digitized,” he writes in his 2018 book AI
Superpowers: China, Silicon Valley, and the New World Order.
That may be changing. The extensive data collection
by viral Chinese face-swapping app Zao provoked outrage not only in the
West, but in China as well, forcing Zao to update its
policy.
And the country with the most data doesn’t
automatically win, anyway. “This is more of a race for human capital than it is
for any particular data source,” says Sellitto of Stanford’s HAI. While
protecting privacy rights may slightly impinge data collection, it helps
attract talent.
The United States has the largest share of
prominent AI researchers, and most of them are foreign born, according to
a study by
the Paulson Institute. The biggest threat to America’s AI leadership
may not be China’s mass of data or the talent developed in other countries, but
newly restrictive immigration policies that make it harder for that talent to
migrate to the U.S. The Partnership on AI, a coalition of businesses and
nonprofits, says that a prohibitive approach to immigration hurts AI
development everywhere. “In the long run, valuing civil liberties is going to
attract the best talent to America by the most innovative people in the world,”
says Khanna. “It allows for freedom of creativity and entrepreneurship in ways
that authoritarian societies don’t.” https://www.fastcompany.com/90402489/5-simple-rules-to-make-ai-a-force-for-
Leading
AI country will be ‘ruler of the world,’ says Putin
"When one party’s drones are destroyed by
drones of another, it will have no other choice but to surrender.”
September 3, 2017
Russian President Vladimir Putin warned Friday (Sept. 1, 2017) that the country
that becomes the leader in developing artificial intelligence will be “the
ruler of the world,” reports the Associated Press.
AI development “raises colossal opportunities and threats that are
difficult to predict now,” Putin said in a lecture to students, warning that
“it would be strongly undesirable if someone wins a monopolist position.”
Future wars will be fought by autonomous drones, Putin suggested, and
“when one party’s drones are destroyed by drones of another, it will have no
other choice but to surrender.”
U.N.
urged to address lethal autonomous weapons
AI experts worldwide are also concerned. On August 20, 116 founders of robotics and artificial intelligence companies from
26 countries, including Elon Musk* and Google DeepMind’s Mustafa Suleyman,
signed an open letter asking the United Nations to “urgently address
the challenge of lethal autonomous weapons (often called ‘killer robots’) and
ban their use internationally.”
“Lethal autonomous weapons threaten to become the third revolution in
warfare,” the letter states. “Once developed, they will permit armed conflict
to be fought at a scale greater than ever, and at timescales faster than humans
can comprehend. These can be weapons of terror, weapons that despots and
terrorists use against innocent populations, and weapons hacked to behave in
undesirable ways. We do not have long to act. Once this Pandora’s box is
opened, it will be hard to close.”
Unfortunately, the box may have already been opened. Three examples:
Russia. In
2014, Dmitry Andreyev of the Russian Strategic
Missile Forces announced that mobile robots would be standing
guard over five ballistic missile installations, New Scientist reported. Armed with a heavy machine gun, this “mobile robotic
complex … can detect and destroy targets, without human involvement.”
In 2016, Russian military equipment manufacturer JSC 766 UPTK announced what appears to be the commercial version:
the Uran-9 multipurpose
unmanned ground combat vehicle. “In autonomous mode, the vehicle can automatically
identify, detect, track and defend [against] enemy targets based on
the pre-programmed path set by the operator,” the company said.
United
States. In a 2016 report, the U.S.
Department of Defense advocated self-organizing “autonomous unmanned” (UA)
swarms of small drones that would assist frontline troops in real time by
surveillance, jamming/spoofing enemy electronics, and autonomously
firing against the enemy.
The authors warned that “autonomy — fueled by advances in artificial
intelligence — has attained a ‘tipping point’ in value. Autonomous capabilities
are increasingly ubiquitous and are readily available to allies and adversaries
alike.” The report advised that the Department of Defense “must take immediate
action to accelerate its exploitation of autonomy while also preparing to
counter autonomy employed by adversaries.”**
South
Korea. Designed initially for the DMZ, Super aEgis
II, a robot-sentry machine gun designed by Dodaam Systems, can identify,
track, and automatically destroy a human target 3 kilometers away, assuming
that capability is turned on.
* “China, Russia, soon all
countries w strong computer science. Competition for AI superiority at national
level most likely cause of WW3 imo.” — Elon Musk tweet 2:33 AM – 4 Sep 2017
** While it doesn’t use AI, the U.S. Navy’s
computer-controlled, radar-guided Phalanx gun
system can automatically detect, track, evaluate, and fire at incoming missiles
and aircraft that it judges to be a threat.
UPDATE Sept. 5, 2017: Added Musk tweet in footnote
related:
- Do We Want Robot Warriors to Decide Who Lives or Dies?
- Why we
really should ban autonomous weapons: a response
- The proposed ban on offensive autonomous weapons is unrealistic and dangerous
Disturbing
video depicts near-future ubiquitous lethal autonomous weapons
The technology
described in the film already exists, says UC Berkeley AI researcher Stuart
Russell
November
18, 2017
Campaign to Stop Killer Robots | Slaughterbots
In
response to growing concerns about autonomous weapons, the Campaign to Stop Killer Robots, a coalition of AI
researchers and advocacy organizations, has released a fictional video
that depicts a disturbing future in which lethal autonomous weapons have become
cheap and ubiquitous worldwide.
UC
Berkeley AI researcher Stuart Russell presented the video
at the United
Nations Convention on Certain Conventional Weapons in Geneva, hosted
by the Campaign to Stop Killer
Robots earlier
this week. Russell, in an appearance at the end of the video, warns that the
technology described in the film already exists* and that the
window to act is closing fast.
Support
for a ban against autonomous weapons has been mounting. On Nov. 2, more
than 200
Canadian scientists and more than 100 Australian scientists in academia and
industry penned open letters to Prime Minister Justin Trudeau and Malcolm
Turnbull urging them to support the ban.
Earlier
this summer, more than 130 leaders of AI companies signed a
letter in
support of this week’s discussions. These letters follow a 2015 open letter released by
the Future of Life Institute and signed by more than 20,000
AI/robotics researchers and others, including Elon Musk and Stephen Hawking.
“Many
of the world’s leading AI researchers worry that if these autonomous weapons
are ever developed, they could dramatically lower the threshold for armed
conflict, ease and cheapen the taking of human life, empower terrorists, and
create global instability,” according to an article published by
the Future of Life Institute, which funded the video.
“The U.S. and other nations have used drones and semi-automated systems to
carry out attacks for several years now, but fully removing a human from the
loop is at odds with international humanitarian and human rights law.”
“The
Campaign to Stop Killer Robots is not trying to stifle innovation in artificial
intelligence and robotics and it does not wish to ban autonomous systems in the
civilian or military world,” explained Noel Sharkey of the International Committee for Robot Arms
Control. Rather we see an urgent need to prevent automation of the critical
functions for selecting targets and applying violent force without human
deliberation and to ensure meaningful human control for every attack.”: http://www.kurzweilai.net/disturbing-video-depicts-near-future-ubiquitous-lethal-autonomous-weapons?utm_source=KurzweilAI+Weekly+Newsletter&utm_campaign=a7e8f35262-UA-946742-1&utm_medium=email&utm_term=0_147a5a48c1-a7e8f35262-282212701
The Quantum
Spy: A Thriller
December
1, 2017 author |David Ignatius
From the
best-selling author of The Director and Body of Lies comes
a thrilling tale of global espionage, state-of-the-art technology, and
unthinkable betrayal.
A
hyper-fast quantum computer is the digital equivalent of a nuclear bomb;
whoever possesses one will be able to shred any encryption and break any code
in existence. The winner of the race to build the world’s first quantum machine
will attain global dominance for generations to come. The question is, who will
cross the finish line first: the U.S. or China?
In
this gripping cyber thriller, the United States’ top-secret quantum research
labs are compromised by a suspected Chinese informant, inciting a mole hunt of
history-altering proportions. CIA officer Harris Chang leads the charge,
pursuing his target from the towering cityscape of Singapore to the lush hills
of the Pacific Northwest, the mountains of Mexico, and beyond. The investigation
is obsessive, destructive, and―above all―uncertain. Do the leaks expose real
secrets, or are they false trails meant to deceive the Chinese? The answer
forces Chang to question everything he thought he knew about loyalty, morality,
and the primacy of truth.
Grounded
in the real-world technological arms race, The Quantum Spy presents
a sophisticated game of cat and mouse cloaked in an exhilarating and visionary
thriller.: http://www.kurzweilai.net/the-quantum-spy-a-thriller?utm_source=KurzweilAI+Weekly+Newsletter&utm_campaign=409682bcab-UA-946742-1&utm_medium=email&utm_term=0_147a5a48c1-409682bcab-28221270
Will AI
enable the third stage of life?
By Max
Tegmark, PhD
August 29, 2017
In his new book Life
3.0: Being Human in the Age of Artificial Intelligence, MIT physicist
and AI researcher Max Tegmark explores the future of technology, life, and
intelligence.
“Tegmark’s
new book is a deeply thoughtful guide to the most important conversation of our
time, about how to create a benevolent future civilization as we merge our
biological thinking with an even greater intelligence of our own creation.” —
Ray Kurzweil, Inventor, Author and Futurist, author of The Singularity
Is Near and How to Create a Mind
In
summary, we can divide the development of life into three stages, distinguished
by life’s ability to design itself:
• Life 1.0 (biological stage): evolves its hardware and software
• Life 2.0 (cultural stage): evolves its hardware, designs much of
its software
• Life 3.0 (technological stage): designs its hardware and
software
After 13.8 billion years of cosmic evolution, development has accelerated
dramatically here on Earth: Life 1.0 arrived about 4 billion years ago,
Life 2.0 (we humans) arrived about a hundred millennia ago, and many
AI researchers think that Life 3.0 may arrive during the coming century,
perhaps even during our lifetime, spawned by progress in AI. What will
happen, and what will this mean for us? That’s the topic of this book.
Artificial Intelligence: An Illustrated History: From Medieval Robots to Neural Networks
by Clifford
A. Pickover
An illustrated
journey through the past, present, and future of artificial intelligence.
From medieval robots and Boolean algebra to facial recognition, artificial
neural networks, and adversarial patches, this fascinating history takes
readers on a vast tour through the world of artificial intelligence.
Award-winning author Clifford A. Pickover (The Math Book, The Physics Book,
Death & the Afterlife) explores the historic and current applications
of AI in such diverse fields as computing, medicine, popular culture,
mythology, and philosophy, and considers the enduring threat to humanity should
AI grow out of control. Across 100 illustrated entries, Pickover provides an
entertaining and informative look into when artificial intelligence began, how
it developed, where it’s going, and what it means for the future of
human-machine interaction.
https://www.goodreads.com/book/show/44443017-artificial-intelligence
Elon Musk wants to enhance us as superhuman cyborgs to deal with superintelligent AI
April 21, 2017
It’s the year 2021. A quadriplegic patient has just had one million “neural lace” microparticles injected into her brain, the world’s first human with an internet communication system using a wireless implanted brain-mind interface — and empowering her as the first superhuman cyborg. …
No, this is not a science-fiction movie plot. It’s the actual first public step — just four years from now — in Tesla CEO Elon Musk’s business plan for his latest new venture, Neuralink. It’s now explained for the first time on Tim Urban’s WaitButWhy blog.
Dealing with the superintelligence existential risk
Such a system would allow for radically improved communication between people, Musk believes. But for Musk, the big concern is AI safety. “AI is obviously going to surpass human intelligence by a lot,” he says. “There’s some risk at that point that something bad happens, something that we can’t control, that humanity can’t control after that point — either a small group of people monopolize AI power, or the AI goes rogue, or something like that.”
“This is what keeps Elon up at night,” says Urban. “He sees it as only a matter of time before superintelligent AI rises up on this planet — and when that happens, he believes that it’s critical that we don’t end up as part of ‘everyone else.’ That’s why, in a future world made up of AI and everyone else, he thinks we have only one good option: To be AI.”
To achieve his, Neuralink CEO Musk has met with more than 1,000 people, narrowing it down initially to eight experts, such as Paul Merolla, who spent the last seven years as the lead chip designer at IBM on their DARPA-funded SyNAPSE program to design neuromorphic (brain-inspired) chips with 5.4 billion transistors (each with 1 million neurons and 256 million synapses), and Dongjin (DJ) Seo, who while at UC Berkeley designed an ultrasonic backscatter system for powering and communicating with implanted bioelectronics called neural dust for recording brain activity.*
Becoming one with AI — a good thing?
Neuralink’s goal its to create a “digital tertiary layer” to augment the brain’s current cortex and limbic layers — a radical high-bandwidth, long-lasting, biocompatible, bidirectional communicative, non-invasively implanted system made up of micron-size (millionth of a meter) particles communicating wirelessly via the cloud and internet to achieve super-fast communication speed and increased bandwidth (carrying more information).
“We’re going to have the choice of either being left behind and being effectively useless or like a pet — you know, like a house cat or something — or eventually figuring out some way to be symbiotic and merge with AI. … A house cat’s a good outcome, by the way.”
But machine intelligence is already vastly superior to human intelligence in specific areas (such as Google’s Alpha Go) and often inexplicable. So how do we know superintelligence has the best interests of humanity in mind?
“Just an engineering problem”
Musk’s answer: “If we achieve tight symbiosis, the AI wouldn’t be ‘other’ — it would be you and with a relationship to your cortex analogous to the relationship your cortex has with your limbic system.” OK, but then how does an inferior intelligence know when it’s achieved full symbiosis with a superior one — or when AI goes rogue?
And what about experts in neuroethics, psychology, law? Musk says it’s just “an engineering problem. … If we can just use engineering to get neurons to talk to computers, we’ll have done our job, and machine learning can do much of the rest.”
However, it’s not clear how we could be assured our brains aren’t hacked, spied on, and controlled by a repressive government or by other humans — especially those with a more recently updated software version or covert cyborg hardware improvements.
In addition, the devices mentioned in WaitButWhy all require some form of neurosurgery, unlike Facebook’s research project to use non-invasive near-infrared light, as shown in this experiment, for example.** And getting implants for non-medical use approved by the FDA will be a challenge, to grossly understate it.
“I think we are about 8 to 10 years away from this being usable by people with no disability,” says Musk, optimistically. However, Musk does not lay out a technology roadmap for going further, as MIT Technology Review notes.
Nonetheless, Neuralink sounds awesome — it should lead to some exciting neuroscience breakthroughs. And Neuralink now has 16 San Francisco job listings here.
* Other experts: Vanessa Tolosa, Lawrence Livermore National Laboratory, one of the world’s foremost researchers on biocompatible materials; Max Hodak, who worked on the development of some groundbreaking BMI technology at Miguel Nicolelis’s lab at Duke University, Ben Rapoport, Neuralink’s neurosurgery expert, with a Ph.D. in Electrical Engineering and Computer Science from MIT; Tim Hanson, UC Berkeley post-doc and expert in flexible Electrodes for Stable, Minimally-Invasive Neural Recording; Flip Sabes, professor, UCSF School of Medicine expert in cortical physiology, computational and theoretical modeling, and human psychophysics and physiology; and Tim Gardner, Associate Professor of Biology at Boston University, whose lab works on implanting BMIs in birds, to study “how complex songs are assembled from elementary neural units” and learn about “the relationships between patterns of neural activity on different time scales.”
** This binary experiment and the binary Brain-to-brain (B2B) internet communication system mentioned above are the equivalents of the first binary (dot–dash) telegraph message, sent May 24, 1844: ”What hath God wrought?”
May 22, 2017
When AI improves human performance instead of taking over
It’s not about artificial intelligence (AI) taking over — it’s about AI
improving human performance, a new study by Yale University researchers
has shown.
“Much of the current conversation about artificial intelligence has to do
with whether AI is a substitute for human beings. We believe the conversation
should be about AI as a complement to human beings,” said Nicholas Christakis, Yale
University co-director of the Yale Institute for Network Science
(YINS) and senior author of a study by Yale Institute for Network
Science.*
AI doesn’t even have to be super-sophisticated to make a difference in
people’s lives; even “dumb AI” can help human groups, based on the study,
which appears in the May 18, 2017 edition of the journal Nature.
How bots can boost human performance
In a series of experiments using teams of human players and autonomous
software agents (“bots”), the bots boosted the performance of human groups and
the individual players, the researchers found.
The experiment design involved an online color-coordination game that
required groups of people to coordinate their actions for a collective goal.
The collective goal was for every node to have a color different than all of
its neighbor nodes. The subjects were paid a US$2 show-up fee and a declining
bonus of up to US$3 depending on the speed of reaching a global solution to the
coordination problem (in which every player in a group had chosen a different
color from their connected neighbors). When they did not reach a global
solution within 5 min, the game was stopped and the subjects earned no bonus.
The human players also interacted with anonymous bots that were programmed
with three levels of behavioral randomness — meaning the AI bots sometimes
deliberately made mistakes (introduced “noise”). In addition, sometimes the
bots were placed in different parts of the social network to try different
strategies.
The result: The bots reduced the median time for groups to solve problems
by 55.6%. The experiment also showed a cascade effect: People whose performance
improved when working with the bots then influenced other human players to
raise their game. More than 4,000 people participated in the experiment, which
used Yale-developed software called breadboard.
The findings have implications for a variety of situations in which people
interact with AI technology, according to the researchers. Examples include
human drivers who share roadways with autonomous cars and operations in which
human soldiers work in tandem with AI.
“There are many ways in which the future is going to be like this,”
Christakis said. “The bots can help humans to help themselves.”
Practical business AI tools
One example: Salesforce CEO Marc Benioff uses a bot called Einstein to help him
run his company, Business Intelligence reported Thursday (May
18, 2017).
“Powered by advanced machine learning, deep learning, predictive analytics,
natural language processing and smart data discovery, Einstein’s models will be
automatically customised for every single customer,” according to the
Salesforce blog. “It will learn, self-tune and get smarter with every
interaction and additional piece of data. And most importantly, Einstein’s
intelligence will be embedded within the context of business, automatically
discovering relevant insights, predicting future behavior, proactively
recommending best next actions and even automating tasks.”
Benioff says he also uses a version called Einstein Guidance for
forecasting and modeling. It even helps end internal politics at executive
meetings, calling out under-performing executives.
“AI is the next platform. All future apps for all companies will be built
on AI,” Benioff predicts.
* Christakis is a professor of sociology, ecology & evolutionary
biology, biomedical engineering, and medicine at Yale. Grants from the Robert
Wood Johnson Foundation and the National Institute of Social Sciences supported
the research.
Abstract of Locally noisy autonomous agents improve global human
coordination in network experiments
Coordination in groups faces a sub-optimization problem and theory
suggests that some randomness may help to achieve global optima. Here we
performed experiments involving a networked colour coordination game in
which groups of humans interacted with autonomous software agents (known as
bots). Subjects (n = 4,000) were embedded in networks (n = 230)
of 20 nodes, to which we sometimes added 3 bots. The bots were programmed with
varying levels of behavioural randomness and different geodesic locations. We
show that bots acting with small levels of random noise and placed in central
locations meaningfully improve the collective performance of human groups,
accelerating the median solution time by 55.6%. This is especially the case
when the coordination problem is hard. Behavioural randomness worked not only
by making the task of humans to whom the bots were connected easier, but also
by affecting the gameplay of the humans among themselves and hence creating
further cascades of benefit in global coordination in these heterogeneous
systems.
references:
Topics: AI/Robotics
The Guardian | God in the Machine: my strange journey into transhumanism
Essay on the future of human
spirituality inspired by Ray Kurzweil's writings.
May 9, 2017
Meghan O’Gieblyn
date: April 18, 2017
date: April 18, 2017
1. | Humanity will change the nature of mortality in our
post-biological future
I first read Ray Kurzweil’s book The Age of Spiritual Machines in
2006. Kurzweil writes, “The 21st century will be different. The human species,
along with the computational technology it created, will be able to solve age
old problems. It will be in a position to change the nature of mortality in a
post-biological future.”
Ray Kurzweil is one of the first major thinkers to bring these ideas to
mainstream — and legitimize them for a wide audience. His ascent in 2012 to a
director of engineering position at Google, heralded for many, a symbolic
merger between the philosophy called transhumanism and the clout of major
technological enterprise. By 2045, Kurzweil predicts technology will be
inside our bodies. At that moment, the arc of progress will curve up into a
vertical line.
2. | Ray Kurzweil writes that humans will be transformed in
“spiritual machines”
Within months of encountering Ray Kurzweil’s book, I became totally
immersed in transhumanist philosophy. I researched topics like nanotechnology
& brain-computer interfaces. I wanted to know if transhumanist ideas
were compatible with Christian eschatology. Was it possible that tech could be
how humanity achieves immortality? At bible school, I studied a branch of
theology that divides all history into successive stages by which god reveals
truth.
Like the theologians at my school, Ray Kurzweil — a leading proponent of
transhumanist philosophy — has his own historical narrative. In his book he
divides the evolution of life into successive epochs. We’re living in the 5th
epoch, when human intelligence begins to merge with tech.
Soon we’ll reach singularity, Kurzweil says — the point where humans will
transforme into what he calls “spiritual machines.” We’ll be able to transfer
our minds to computers, letting us live forever. Our bodies will also become
immune to disease & aging. Using tech, humanity will transform earth into a
paradise — then migrate to space, terra-forming other planets.
Alpha Go defeats world’s top Go player. What’s next?
May 28, 2017
What does the research team behind AlphaGo do next after winning the
three-game match Saturday (May 27) against Ke Jie — the world’s top Go player —
at the Future of Go Summit in Wuzhen, China?
“Throw their energy into the next set of grand challenges, developing advanced general algorithms that could one day help scientists as they tackle some of our most complex problems, such as finding new cures for diseases, dramatically reducing energy consumption, or inventing revolutionary new materials,” says DeepMind Technologies CEO Demis Hassabis.
Academic paper, Go teaching tool
But it’s “not the end of our work with the Go community,” he adds. “We plan
to publish one final academic paper later this year that will detail the
extensive set of improvements we made to the algorithms’ efficiency and
potential to be generalised across a broader set of problems.”
High-speed light-based
systems could replace supercomputers for certain ‘deep learning’ calculations
Low power requirements for photons
(instead of electrons) may make deep learning more practical in future
self-driving cars and mobile consumer devices
June 14, 2017
A team of researchers at MIT and elsewhere has developed a new approach
to deep learning systems
— using light instead of electricity, which they say could vastly improve the
speed and efficiency of certain deep-learning computations.
Deep-learning systems are based on artificial neural
networks that mimic the way the brain learns from an
accumulation of examples. They can enable technologies such as face- and
voice-recognition software, or scour vast amounts of medical data to find
patterns that could be useful diagnostically, for example.
But the computations these systems carry out are highly complex and
demanding, even for supercomputers. Traditional computer architectures are not
very efficient for calculations needed for neural-network tasks that involve
repeated multiplications of matrices (arrays
of numbers). These can be computationally intensive for conventional CPUs or
even GPUs.
Programmable nanophotonic processor
Instead, the new approach uses an optical device that the researchers call
a “programmable nanophotonic processor.” Multiple light beams are directed in
such a way that their waves interact with each other, producing interference patterns that “compute” the intended
operation.
The optical chips using this architecture could, in principle, carry out
dense matrix multiplications (the most power-hungry and time-consuming part in
AI algorithms) for learning tasks much faster, compared to conventional
electronic chips. The researchers expect a computational speed enhancement of
at least two orders of magnitude over the state-of-the-art and three orders of
magnitude in power efficiency.
“This chip, once you tune it, can carry out matrix multiplication with, in
principle, zero energy, almost instantly,” says Marin Soljacic, one of the MIT
researchers on the team.
To demonstrate the concept, the team set the programmable nanophotonic
processor to implement a neural network that recognizes four basic vowel
sounds. Even with the prototype system, they were able to achieve a 77 percent
accuracy level, compared to about 90 percent for conventional systems. There
are “no substantial obstacles” to scaling up the system for greater accuracy,
according to Soljacic.
The team says is will still take a lot more time and effort to make this
system useful. However, once the system is scaled up and fully functioning, the
low-power system should find many uses, especially for situations where power
is limited, such as in self-driving cars, drones, and mobile consumer devices.
Other uses include signal processing for data transmission and computer centers.
The research was published Monday (June 12, 2017) in a paper in the journal Nature Photonics (open-access
version available on arXiv).
The team also included researchers at Elenion Technologies of New York and
the Université de Sherbrooke in Quebec. The work was supported by the U.S. Army
Research Office through the Institute for Soldier Nanotechnologies, the
National Science Foundation, and the Air Force Office of Scientific Research.
Abstract of Deep learning with coherent nanophotonic circuits
Artificial neural networks are computational network models inspired by
signal processing in the brain. These models have dramatically improved
performance for many machine-learning tasks, including speech and image
recognition. However, today’s computing hardware is inefficient at implementing
neural networks, in large part because much of it was designed for von Neumann
computing schemes. Significant effort has been made towards developing
electronic architectures tuned to implement artificial neural networks that
exhibit improved computational speed and accuracy. Here, we propose a new
architecture for a fully optical neural network that, in principle, could offer
an enhancement in computational speed and power efficiency over
state-of-the-art electronics for conventional inference tasks. We
experimentally demonstrate the essential part of the concept using a
programmable nanophotonic processor featuring a cascaded array of 56
programmable Mach–Zehnder interferometers in a silicon photonic integrated circuit
and show its utility for vowel recognition.
references:
· Next-generation
computer chip with two heads
November 5, 2020
Summary:
Engineers have developed a
computer chip that combines two functions - logic operations and data storage -
into a single architecture, paving the way to more efficient devices. Their
technology is particularly promising for applications relying on artificial
intelligence…:
https://www.sciencedaily.com/releases/2020/11/201105112954.htm
How
Artificial Intelligence will Transform IT Operations and DevOps
- Published on May 11, 2017
- Featured in: Big Data, Cloud Computing, Information
Technology
Director Adversitement•Helping Data Driven Companies
Generating Success•Top10 Big Data, Data Science, IoT, BI Influencer
To state that DevOps and IT operations teams will face
new challenges in the coming years sounds a bit redundant, as their core
responsibility is to solve problems and overcome challenges. However, with the
dramatic pace in which the current landscape of processes, technologies, and
tools are changing, it has become quite problematic to cope with it. Moreover,
the pressure business users have been putting on DevOps and IT operations teams
is staggering, demanding that everything should be solved with a tap on an app.
However, at the backend, handling issues is a different ball game; the users
can’t even imagine how difficult it is to find a problem and solve it.
One of the biggest challenges IT operations and DevOps
teams face nowadays is being able to pinpoint the small yet potentially harmful
issues in large streams of Big Data being logged in their environment. Put
simply, it is just like finding a needle in the haystack.
If you work in the IT department of a company with
online presence that boasts 24/7 availability, here is a scenario that may
sound familiar to you. Assume that you get a call in the middle of the night
from an angry customer or your boss complaining about a failed credit card
transaction or an application crash. You go to your laptop right away and open
the log management system. You see there are a more than a hundred thousand
messages logged at the set timeframe – a data set impossible for a human being
to review line by line.
So what do you do in such a situation?
It is the story of every IT operations and DevOps
professional; they spend many sleepless nights, navigating through the sea of
log entries to find critical events that triggered a specific event. This is
where real-time and centralized log analytics come to the rescue. It helps them
in understanding the essential aspects of their log data, and easily identify
the main issues. With this, the troubleshooting process becomes a walk in the
park, making it shorter and more effective, as well as enabling experts to
predict the future problems.
AI and Its Effect on IT Operations
and DevOps
While Artificial Intelligence (AI) used to be the
buzzword a few decades ago, it is now being commonly applied across different
industries for a diverse range of purposes. Combining big data, AI, and human
domain knowledge, technologists and scientists have become able to create
astounding breakthroughs and opportunities, which used to be possible in
science fiction novels and movies only.
As IT operations become agile and dynamic, they are
also getting immensely complex. The human mind is no longer capable of keeping
up with the velocity, volume, and variety of Big Data streaming through daily
operations, making AI a powerful and essential tool for optimizing the
analyzing and decision-making processes. AI helps in filling the gaps between
humans and Big Data, giving them the required operational intelligence and
speed to significantly waive off the burden of troubleshooting and real-time
decision-making.
Addressing the Elephant in the Room
– How AI can Help
In all the above situations, one thing is common;
these companies need a solution – as discussed in the beginning – that helps IT
and DevOps teams to quickly find problems in the mountain of log data entries.
To identify that single log entry putting cracks in the environment and
crashing your applications, wouldn’t it be easy if you just knew what kind of
error you are looking for to filter your log data? Of course, it would cut down
the amount of work by half.
One solution can be to have a platform that has
collected data from the internet about all kinds of related incidents, observed
how people using similar setups resolved them in their systems, and scanned
through your system to identify the potential problems. One way to achieve this
is to design a system that mimics how a user investigates, monitors, and
troubleshoots events, and allows it to develop an understating how humans
interact with data instead of trying to analyze the data itself. For example,
this technology can be similar to Amazon’s product recommendation system and
Google’s PageRank algorithm, but it will be focused on log data.
Introducing Cognitive
Insights
A recent technology implements a solution as
envisioned by this post. The technology - which has been generating quite a lot
of buzz lately- is called Cognitive Insights. This groundbreaking technology
uses machine-learning algorithms to match human domain knowledge with log data,
along with open source repositories, discussion forums, and social thread.
Using all this information, it makes a data reservoir of relevant insights that
may contain solutions to a wide range of critical issues, faced by IT
operations and DevOps teams on a daily basis.
The Real-Time Obstacles
DevOps engineers, IT Operations managers, CTOs, VP
engineering, and CISO face numerous challenges, which can be mitigated
effectively by integrating AI in log analysis and related operations. While
there are several applications of Cognitive Insights, the two main use cases
are:
- Security
Distributed Denial of Service (DDoS) attacks are
increasingly becoming common. What used to be just limited to governments,
high-profile websites, and multinational organizations is now targeting
prominent individuals, SMBs and mid-sized enterprises.
To ward off such attacks, having a centralized logging
architecture to identify suspicious activities and pinpoint the potential
threats from thousands of entries is essential. For this, anti-DDoS mitigation
through Cognitive Insights has been proven to be highly effective. Leading
names, such as Dyn and British Airways, that sustained significant damage from
DDoS attacks in the past now have a full-fledge, ELK-based anti-DDoS mitigation
strategy in place to keep hackers at bay and secure their operations from any
future attacks.
IT Operations
Wouldn’t it be great to have all your logs compiled
into a single place, with each entry carefully monitored and registered? Well,
certainly. You will be able to view the process flow clearly and execute
queries pertaining to the logs from different applications all from one place,
hence dramatically increasing the efficiency of your IT operations. To solve
one of the biggest challenges IT operations and DevOps teams face is being able
to pinpoint the small yet potentially harmful issues in large streams of log
data in their environment. This is precisely what Cognitive Insights does.
Since the core of this program is based on the ELK stack, it sorts and
simplifies the data and makes it easy to have clear picture of your IT
operations. Asurion and Performance Gateway are perfect examples that have
leveraged from Cognitive Insights and taken their IT game up a notch.
The Good AI Integration can Yield
Using AI driven log analytics systems, it becomes
considerably easy to find the needle in the haystack, and efficiently solve
issues. Such a system will have a considerable impact on management and
operations of the entire organization. Like the problems of companies discussed
above in this blog, integrating AI with log management system will benefit in:
- Improved customer success
- Monitoring and customer support
- Risk reduction and resource optimization
- Maximize efficiency by making logging data
accessible
In other words, Cognitive Insights and other similar
systems can be of great help in data log management and troubleshooting.
Rent-A-Center (RAC) is a Texas-based, Fortune 1000
company that offers a wide range of rent-to-own products and services. It has
over 3000 stores and 2000 kiosks spread across Mexico, Puerto Rico, Canada, and
United States. The company tried integrating two different ELK stacks, but
handling 100GB data every day was too much of a hassle, not to mention the
exorbitant cost and time spent every day for disk management, memory tuning,
additional data input capabilities, and other technical issues. RAC
transitioned to Cognitive Insights, which gave them the confidence that they
will be able to detect future anomalies and made it quite easily to scale the
constantly growing volume of data. They benefitted from a dedicated IT team
managing on-premise and off-premise ELK stacks.
The Role of Open Source in Data Log
Management
Many reputed vendors are proactively researching and
testing AI in different avenues to enhance the efficiency of data log
management systems. Some of the vendors are:
There is no surprise in the fact that ELK is fast
becoming part of the trend, and more and more vendors are offering logging
solutions. This is because it has become a great way for companies to install a
setup without incurring a staggering upfront cost. It also allows for some
basic graphing and searching capabilities, and in order for the organizations
to recognize the issues in their haystack of log data, they can opt for latest
technologies, like Cognitive Insights, to quickly find the needle and eliminate
the main problems.
Introduction
Concern about a “ jobless future” has never been
greater. Seemingly every day, an academic, researcher or technology leader
suggests that in a world of automation and artificial intelligence (AI),
workers will increasingly be a surplus to what businesses need – or as Stanford
University’s Jerry Kaplan says in his best-selling book, it won’t be long
before “humans need not apply.”1 The concerns are understandable. AI – long
academic theory and Hollywood plotline – is becoming “real” at an astonishing
pace and finding its way into more and more aspects of work, rest and play. AI
is now being used to read X-rays and MRIs. It’s at the heart of stock trading.
Chat with Siri or Alexa, and you’re using AI. Soon, AI will be found in every
job, profession and industry around the world. When machines do everything,
lots of people wonder what will we do? What work will be left for people? How
will we make a living when machines are cheaper, faster and smarter than we are
– machines that don’t take breaks or vacations, don’t get sick and don’t care
about chatting with their colleagues about last night’s game? For many people,
the future of work looks like a bleak place, full of temporary jobs (a “gig”
economy), minimum wage labor and a ruling technocracy safely hidden away in
their gated communities and their circular living machines. Although plausible,
this vision of the future is not one we share. Our vision is quite different –
and much more optimistic. Our vision is based on a different reading of the
trends and the facts; a different interpretation of how change occurs and how
humans evolve. Our view of the future of work is based on the following
principles:… https://www.cognizant.com/whitepapers/21-jobs-of-the-future-a-guide-to-getting-and-staying-employed-over-the-next-10-years-codex3049.pdf
Are We Ready for Quantum Computers?
Hardware
hasn’t caught up with theory, but we’re already lining up many previously
intractable problems for when it does
- By Rolando
Somma on March 13, 2020
A
recent paper by
Google claiming that a quantum computer performed a specific calculation that
would choke even the world’s fastest classical supercomputer has raised many
more questions than it answered. Chief among them is this: When full-fledged
quantum computers arrive, will we be ready?
Google
achieved this milestone against the backdrop of a more sobering reality: Even
the best gate-based quantum computers today can only muster around 50 qubits. A
qubit, or quantum bit, is the basic piece of information in quantum computing,
analogous to a bit in classical computing but so much more.
Gate-based
quantum computers operate using logic gates but, in contrast with classical
computers, they exploit inherent properties of quantum mechanics such as
superposition, interference and entanglement. Current quantum computers are so
noisy and error-prone that the information in its quantum state is lost within
tens of microseconds through a mechanism called decoherence and through faulty
gates.
Still,
researchers are making demonstrable, if slow, progress toward more usable
qubits. Perhaps in 10 years, or 20, we’ll reach the goal of reliable,
large-scale, error-tolerant quantum computers that can solve a wide range of
useful problems.
When
that day comes, what should we do with them?
We’ve
had decades to prepare. In the early 1980s, the American physicist Paul Benioff
published a paper demonstrating
that a quantum-mechanical model of a Turing machine—a computer—was
theoretically possible. Around the same time, Richard Feynman argued that
simulating quantum systems at any useful scale on classical computers would
always be impossible because the problem would get far, far too big: the
required memory and time would increase exponentially with the volume of the
quantum system. On a quantum computer, the required resources would scale up
far less radically.
Feynman
really launched the field of quantum computing when he suggested that the best
way to study quantum systems was to simulate them on quantum computers.
Simulating quantum physics is the app for quantum computers.
They’re not going to be helping you stream video on your smartphone. If large,
fault-tolerant quantum computers can be built, they will enable us to probe the
strange world of quantum mechanics to unprecedented depths. It follows
different rules than the world we observe in our everyday lives and yet
underpins everything.
On
a big enough quantum computer, we could simulate quantum field theories to
study the most fundamental nature of the universe. In chemistry and nanoscale
research, where quantum effects dominate, we could investigate the basic
properties of materials and design new ones to understand mechanisms such as
unconventional superconductivity. We could simulate and understand new chemical
reactions and new compounds, which could aid in drug discovery.
By
diving deep into mathematics and information theory, we already have developed
many theoretical tools to do these things, and the algorithms are farther along
than the technology to build the actual machines. It all starts with a
theoretical model of the quantum computer, which establishes how it will
harness quantum mechanics to perform a useful computation. Researchers write
quantum algorithms to perform a task or solve a problem using that model. These
are basically a sequence of quantum gates together with a measurement of the
quantum state that provides the desired classical information.
So,
for instance, Grover’s
algorithm shows a way to perform faster searches. Shor’s
algorithm has proved that large quantum computers will one day be able
to break computer security systems based on RSA, a method widely used to
protect, for instance, e-mail and financial websites worldwide.
In
my research, my colleagues and Ihave demonstrated
very efficient algorithms to perform useful computations and study
physical systems. We have also demonstrated one of the methods in one of
the first small-scale quantum simulations ever done of a system of electrons,
in a nuclear
magnetic resonance quantum information processor. Others have also followed
up on our work and recently
simulated simple quantum field theories on the noisy intermediate scale
quantum computers available today and in laboratory experiments.
Brain-computer-interface training helps
tetraplegics win avatar race
When humans
actively participate with AI in improving performance
May 14, 2018
Noninvasive brain–computer
interface (BCI) systems can restore functions lost to disability — allowing for
spontaneous, direct brain control of external devices without the risks
associated with surgical implantation of neural interfaces. But as
machine-learning algorithms have become faster and more powerful, researchers
have mostly focused on increasing performance by optimizing pattern-recognition
algorithms.
But what about letting
patients actively participate with AI in improving performance?
To test that idea,
researchers at the École Polytechnique Fédérale de Lausanne (EPFL), based in
Geneva, Switzerland, conducted research using “mutual learning” between
computer and humans — two severely impaired (tetraplegic) participants with
chronic spinal cord injury. The goal: win a live virtual racing game at an
international event.
Controlling a racing-game avatar using a
BCI
The participants were
trained to improve control of an avatar (a person-substitute shown on a
computer screen) in a virtual racing game. The experiment used a brain-computer
interface (BCI), which uses electrodes on the head to pick up control signals
from a person’s brain.
Each participant (called a
“pilot”) controlled an on-screen avatar in a three-part race. This required
mastery of separate commands for spinning, jumping, sliding, and walking
without stumbling.
After training for several
months, in Oct. 8, 2016, the two pilots participated (on the “Brain Tweakers”
team) in Cybathlon in
Zurich, Switzerland — the first international para-Olympics for disabled
individuals in control of bionic assistive technology.*
The BCI-based race consisted
of four brain-controlled avatars competing in a virtual racing game called
“Brain Runners.” To accelerate each pilot’s avatar, they had to issue up to
three mental commands (or intentional idling) on corresponding color-coded
track segments.
Maximizing BCI performance
by humanizing mutual learning
The two participants in the EPFL research had the best
three times overall in the competition.
One of those pilots won the gold medal and the other held the tournament
record.
The researchers believe that
with the mutual-learning approach, they have “maximized the chances for human
learning by infrequent recalibration of the computer, leaving time for the
human to better learn how to control the sensorimotor rhythms that would most
efficiently evoke the desired avatar movement. Our results showcase strong and
continuous learning effects at all targeted levels — machine, subject, and
application — with both [participants] over a longitudinal study lasting
several months,” the researchers conclude.
* At Cybathlon,
each team comprised a pilot together with scientists and technology providers
of the functional and assistive devices used, which can be prototypes developed
by research labs or companies, or commercially available products. That also
makes Cybathlon a competition between companies and research
laboratories. The next Cybathlon will be held in Zurich in 2020.
Don’t Regulate Artificial Intelligence:
Starve It
By William Davidow, Michael
S. Malone on May 4, 2020
Artificial intelligence is
still in its infancy. But it may well prove to be the most powerful technology
ever invented. It has the potential to improve health, supercharge intellects,
multiply productivity, save the environment and enhance both freedom and
democracy.
But as that intelligence
continues to climb, the danger from using AI in an irresponsible way also
brings the potential for AI to become a social and cultural H-bomb. It’s a
technology that can deprive us of our liberty, power autocracies and genocides,
program our behavior, turn us into human machines and, ultimately, turn us into
slaves. Therefore, we must be very careful about the ascendance of AI; we don’t
dare make a mistake. And our best defense may be to put AI on an extreme diet.
We already know certain
threatening attributes of AI. For one thing, the progress of this technology
has been, and will continue to be, shockingly quick. Many people were likely
stunned to read recently the announcement by Microsoft that AI was proving to
be better at reading X-rays than trained radiologists. Most newspaper readers
don’t realize how much of their daily paper is now written by AI. That wasn’t
supposed to happen; robots were supposed to supplant manual labor jobs, not
professional brainwork. Yet here we are: AI is quickly gobbling up entire
professions—and those jobs will never come back.
We also are getting closer
to creating machines capable of artificial general intelligence—that is,
machines as intelligent as humans. We may never get all of the way to actual
consciousness, but in terms of processing power, inference, metaphor and even
acquired wisdom, it is easy to imagine AI surpassing humanity. More than 20
years ago, chess master Gary Kasparov playing IBM’s supercomputer Deep Blue,
sensed a mind on the other side of the board. Today, there are hundreds of
thousands of computers in use around the world that are more powerful than Deep
Blue—and that doesn’t include millions of personal computers with access to the
cloud.
We also know that profit
motives and the will to power and control have already driven the rapid growth
of vast libraries of antisocial applications. We need look no farther than the
use of facial recognition and other AI techniques by the government of China to
control the behavior of its citizens to see one such trajectory. That country’s
Social Credit System monitors the behavior of millions of its citizens,
rewarding them for what the government judges to be “good” behavior—and
punishes them for “bad” behavior—by expanding or limiting their access to the
institutions of daily life. Those being punished often do not even know that
their lives are being circumscribed. They are simply not offered access to
locations, promotions, entertainment and services enjoyed by their neighbors.
Meanwhile, here in the free
world the most worrisome threat is the use of AI by industry to exploit us—and
of special interest groups to build and manipulate affinity groups to
increasingly polarize society. The latter activity is particularly egregious in
election years like this one. We are also concerned about the use by law
enforcement, the IRS and regulators to better surveil people who might commit
crimes, evade taxes and commit other transgressive acts. Some of this is
necessary—but without guardrails it can lead to a police state.
Sound extreme? Consider that
already all of us are being detained against our wills, often even against our
knowledge, in what have been called “algorithmic
prisons.” We do not know who sentenced us to them or even the terms
of that sentence. What we do know is that based upon a decision made by some AI
system about our behavior (such as a low credit rating), our choices are being
limited. Predetermination is being made about the information we see: whether a
company will look at our resume, or whether we are eligible for a home loan at
a favorable rate, if we can rent a certain apartment, how much we must pay for
car insurance (our driving quality monitored by new devices attached to our
engine computers), whether we will get into the college of our choice and
whether police should closely monitor our behavior.
Looking ahead, we can be
certain that such monitoring will grow. We know as well that AI will be used by
groups to recruit members and influence their opinions, and by foreign
governments to influence elections. We can also be certain that as AI tools
become more powerful and as the Internet of Things grows, the arsenal of the
virtual weapons will become more commercially—and socially—deadly.
We need to act. The problem
is that, even now, it will be hard to get the horse back into the barn. The
alarm about the growing power of AI already has led to warnings from the likes
of Stephen Hawking and Elon Musk. But it is hard to figure out what to do
legislatively. We haven’t seen any proposals that would have a broad impact,
without crushing the enormous potential advantages of AI.
Europeans now have the
“right to explanation,” which requires a humanly readable justification for all
decisions rendered by AI systems. Certainly, that transparency is desirable,
but it is not clear how much good it will do. After all, AI systems are in
constant flux. So, any actions taken based on the discovery of an injustice
will be like shaping water. AI will just adopt a different shape.
We think a better approach
is to make AI less powerful. That is, not to control artificial
intelligence, but to put it on an extreme diet. And what does AI consume? Our
personal information.
If AI systems and the
algorithms in charge of “virtual prisons” cannot get their hands on this
personal information, cannot indulge their insatiable hunger for this data,
they necessarily will become much less intrusive and powerful.
How do we choke down the
flow of this personal information? One obvious way is to give individuals
ownership of their private data. Today, each of us is surrounded by a penumbra
of data that we continuously generate. And that body of data is a free target
for anyone who wishes to capture and monetize it. Why not, rather than letting
that information flow directly into the servers of the world, instead store it
in the equivalent of a safe deposit box at an information fiduciary like
Equifax? Once it is safely there, the consumer could then decide who gets
access to that data.
For example, suppose a
consumer wants to get a loan, he or she could release the relevant information
to a credit provider—who in turn would have the right to use that information
for that one instance. If that consumer wants to get free service from, say,
Facebook, he or she could provide the company with relevant information for
that application alone. If the government needs access to that information to
catch a terrorist, it will need to get a search warrant. (Another nice feature
of such a system would be that the consumer would only have to go to one place
to check the accuracy of the information on file.)
Human society existed for
millennia before AI systems had unlimited knowledge about each of us. And it
will continue to exist, even if we limit that knowledge by starving our
machines of that personal information. AI will still be able to make the
economy more efficient, create medical advances, reduce traffic and create more
effective regulations to ensure the health of the environment. What it will be
less able to do is threaten human autonomy, liberty and pursuit of happiness.
In the case of AI, lean will
mean less mean. It’s time to put artificial intelligence on a data diet.
Elon Musk Preaches Dangers of AI to NGA, AI is the Next Nuclear Bomb of
Warfare, and More AI News This Week
July 20, 2017
Robert Light Senior Research Specialist at G2 Crowd
ELON MUSK AND DANGERS OF AI
Elon Musk is one of the most outspoken proponents
of regulating AI, and earlier this week he got an opportunity to voice his opinions in front of the National Governors
Association. As the CEO of Tesla, Musk has vested business interest in the
use of AI, yet still does his best to inform others of the dangers he believes
AI poses. “‘I have exposure to the very cutting-edge AI, and I think people
should be really concerned about it’” Musk said during his speech at the
National Governors Association Saturday.
Musk believes that regulation must be set up
immediately, otherwise the results may be detrimental. “AI is a rare
case where we need to be proactive about regulation instead of reactive.
Because I think by the time we are reactive in AI regulation, it’s too late.”
Alluding to reactive regulation examples, Musk
said, “AI is a fundamental risk to the existence of human civilisation, in a
way that car accidents, airplane crashes, faulty drugs or bad food were not.
They were harmful to a set of individuals in society, but they were not harmful
to society as a whole … AI is a fundamental existential risk for human
civilisation, and I don’t think people fully appreciate that.”
However, not everyone agrees with Musk’s
sentiments surrounding the dangers of AI, including the founding director of MIT’s Computer Science and Artificial
Intelligence Lab, Rodney Brooks. In an interview earlier this week, Brooks
was asked to comment on Musk’s lecture on the need for regulations. “So you’re
going to regulate now. If you’re going to have a regulation now, either it
applies to something and changes something in the world, or it doesn’t apply to
anything. If it doesn’t apply to anything, what the hell do you have the
regulation for? Tell me, what behavior do you want to change, Elon? By the way,
let’s talk about regulation on self-driving Teslas, because that’s a real
issue.”
Brooks is not alone in his disagreements with
Musk. Other AI researchers publically defended the advancement of AI for
the good of humanity on social media. Regardless of who is correct, it is
entertaining to listen to smart people argue, especially when it gets personal.
This originally appeared in G2 Crowd's AI Digest. Subscribe to
receive the same weekly AI news directly to your inbox.
HOW AI WILL IMPACT WARFARE
Now for a topic that is much scarier than an
impending AI apocalypse: Governments are utilizing AI in warfare! Greg Allen and Taniel Chan recently released a research reportdiscussing
how artificial intelligence will help national security.
According to Wired, “One of its conclusions is that the
impact of technologies such as autonomous robots on war and international
relations could rival that of nuclear weapons … It lays out why technologies
like drones with bird-like agility, robot hackers and software that generates
photo-real fake video are on track to make the American military and its rivals
much more powerful.”
Unlike an AI apocalypse, these threats to society
feel much more tangible and realistic. The Kalashnikov Group is already working on military AI’s using
neural networks. These robots, in theory, would be able to determine the enemy
and execute them. Feels eerily similar to Robocop.
While these technologies are not an immediate
concern, they will be sometime in the future, making them frightening
nonetheless.
Who can save Humanity from Superintelligence?
Dates: April 29, 2017
Location: London, UK
Location: London, UK
In this presentation, Tony Czarnecki, Managing Partner of Sustensis, will
share his views on how Humanity could be saved from its biggest existential
risk, Superintelligence.
The presentation will cover four overlapping crises Humanity faces today –
crises in the domains of politics, economics, society, and existential risk.
The presentation will also provide a vision of a possible solution, with a
reformed European Union becoming the core of a new supranational organization having
the best chance to tackle these problems.
About the crises:
The world faces a series of existential risks. When combined, the chance of
one of these risks materializing in just 20 years is at least 5%. We already
had one such “near miss” that could have annihilated the entire civilization.
That was the Cuban crisis in October 1962, which almost started a global
nuclear war. Today, the biggest risk facing our civilization and humanity
isSuperintelligence.
Additionally, mainly due to the advancement in technology, the world is
changing at almost an exponential pace. That means that change, not just in
technology but also in political or social domains, which might previously have
taken a decade to produce a significant effect, can now happen in just a year
or two. No wonder that people, even in the most developed countries, cannot
absorb the pace of change that happens simultaneously in so many domains of our
lives. That’s why emotions have overtaken reason.
People are voting in various elections and referenda against the status
quo, not really knowing what the problem is, even less what could be the
solution. Even if some politicians know what the overall, usually unpleasant
solutions could be, they are unlikely to share that with their own electorate
because they would be deselected in the next election. The vicious circle
continues but at an increasingly faster pace.
The crises that we are experiencing right now lie in four domains:
·
Existential survival – the biggest crisis because it is barely visible
·
Political – the crisis of democracy
·
Economic – the crisis of capitalism
·
Social – the crisis of wealth distribution where the wealthy become
wealthier even faster.
At the same time, anyone wanting to improve the situation faces three
problems:
·
Existential risks require fast action, while the world’s organisations act
very slowly
·
People want more freedom and more control, while we need to give up some of
our freedoms and national sovereignty for the greater good of civilisation and
humanity
·
Most people can’t see beyond tomorrow and act emotionally, while we need to
see the big picture and act rationally.
Therefore, anybody that sees the need for the world to take urgent action
faces a formidable task of proposing pragmatic, fast and very radical changes
in the ways the world is governed.
The key problem is that the world cannot put faith in the United Nations –
the organisation that should by default be responsible for leading the humanity
through this most difficult period of existential threats. Neither do we have
time to build such an organisation from scratch. But more importantly, we
cannot reasonably expect that all major blocs – the USA, China, Russia or the
EU – would suddenly replace their own set of values and interests with a
unified set of new human values and responsibilities.
Therefore, the only plausible solution is to rely on an organisation
emerging from a deeply transformed existing organisation, such as the European
Union and/or NATO, which would lead humanity to a new era, where we may be
living side by side with Superintelligence. This would also mean that in the
transition period, this enlarged organization would have to co-exist with
China, Russia, Saudi Arabia and many other countries with deeply different
values and interests.
About the speaker:
Tony Czarnecki is
Managing Partner of Sustensis, a management consultancy that has specialized
for over 20 years in the area of long-term sustainable growth. The objective of
Sustensis is to help companies make a gradual transition from a short-term to a
long-term business growth.
In recent years, Tony has applied his experience in “long-termism” to find
solutions for the crises facing our civilisation and humanity – the
subject that has been the focus of London Futurists.
Tech world debate on robots and jobs heats up
March 26, 2017 by Rob Lever
Advances in artificial intelligence and robotics are heightening concerns
about automation replacing a growing number of occupations
Are robots coming for your job?
Although technology has long affected the labor force, recent advances in artificial intelligence and
robotics are heightening concerns about automation replacing a growing number
of occupations, including highly skilled or "knowledge-based" jobs.
Just a few examples: self-driving technology may eliminate the need for
taxi, Uber and truck drivers, algorithms are playing a growing role in
journalism, robots are informing consumers as mall greeters, and medicine is
adapting robotic surgery and artificial intelligence to detect cancer and heart
conditions.
Of 700 occupations in the United States, 47 percent are at "high
risk" from automation, an Oxford University study concluded in 2013.
A McKinsey study released this year offered a similar view, saying
"about half" of activities in the world's workforce "could
potentially be automated by adapting currently demonstrated technologies."
Still, McKinsey researchers offered a caveat, saying that only around five
percent of jobs can be "fully automated."
Another report, by PwC this month, concluded that around a third of jobs in
the United States, Germany and Britain could be eliminated by automation by the
early 2030s, with the losses concentrated in transportation and storage,
manufacturing, and wholesale and retail trade.
But experts warn that such studies may fail to grasp the full extent of the
risks to the working population.
"The studies are underestimating the impact of technology—some 80 to
90 percent of jobs will be eliminated in the next 10 to 15 years," said
Vivek Wadhwa, a tech entrepreneur and faculty member at Carnegie Mellon
University in Silicon Valley.
Dire consequences
"Artificial intelligence is moving a lot faster than anyone had
expected," said Wadhwa, who is co-author of a forthcoming book on the
topic. "Alexa (Amazon's home hub) and Google Home are getting amazingly
intelligent very fast. Microsoft and Google have demonstrated that AI can
understand human speech better than humans can."
Wadhwa calls the driverless car a "metaphor" for the future of
labor and a sign of a major shift.
Warnings of dire social consequences from automation have also come from
the likes of the physicist Stephen Hawking and tech entrepreneur Elon Musk,
among others.
Hebrew University of Jerusalem historian Yuval Harari writes in his 2017
book, "Homo Deus: A Brief History of Tomorrow" that technology will
lead to "superfluous people" as "intelligent non-conscious
algorithms" improve.
"As algorithms push humans out of the job market," he writes,
"wealth and power might become concentrated in the hands of the tiny elite
that owns the all-powerful algorithms, creating unprecedented social and political
inequality."
Harari points to the Oxford study, estimating a high probability of job
loss to automation—cashiers (97 percent), paralegals (94 percent), bakers (89
percent) and bartenders (77 percent), for example.
Others disagree.
Boston University economist and researcher James Bessen dismisses alarmist
predictions, contending that advances in technology generally lead to more
jobs, even if the nature of work changes.
His research found that the proliferation of ATM machines did not decrease
bank tellers' employment in recent decades, and that automation of textile
mills in the 19th century led to an increase in weaving jobs because it created
more demand.
"Robots can replace humans in certain tasks but don't entirely replace
humans," he said.
But he acknowledged that automation "is destroying a lot of low-skill,
low wage jobs, and the new jobs being created need higher skills."
Former president Barack Obama's council of economic advisors also warned
last year that most jobs paying less than $20 an hour "would come under
pressure from automation."
'Tax the robot'
Although the net impact of robots remains unclear, tech leaders and others
are already debating how to deal with the potential job displacement.
Microsoft founder Bill Gates said last month that he supports a "robot
tax," an idea floated in Europe, including by a socialist presidential
candidate in France.
But Bessen, a former fellow at Harvard's Berkman Center, said taxing robots
could be counterproductive.
"You don't want to be taxing the machines because they enable people
to earn higher wages," he said. "If you tax machines, you will slow
the beneficial side of the process."
Peter Diamandis, chairman of the X Prize Foundation for technical
innovation and founder of the Silicon Valley think-tank Singularity University,
is among those calling for a "universal basic income" to compensate
people for job losses.
Offering income guarantees "will be one of many tools empowering
self-actualization at scale," he said in a blog post, arguing that automation
will allow people "to follow their passions, be more creative."
But Wadhwa says the problems run deeper and will require more creative
solutions.
"A basic income won't solve the social problems of joblessness because
people's identity revolves around our jobs," he said.
"Even if we have enough food and energy, we have to deal with the
social disruption that's coming. We need a much broader discussion."
Bessen says reversing the trends of the past decades, where high-skilled
jobs gain at the expense of others, pose a "big challenge."
"It's entirely possible we can meet the challenge," he said.
"But the evidence in the past 20 years is that things are moving in the
wrong direction."
Read more at: https://phys.org/news/2017-03-tech-world-debate-robots-jobs.html#jCp
Around the halls: What should
the regulation of generative AI look like?
Nicol Turner Lee, Niam Yaraghi, Mark MacCarthy, and Tom Wheeler Friday, June 2, 2023
We are living in a time of unprecedented advancements in generative artificial intelligence (AI), which are AI systems that can generate a wide range of content, such as text or images. The release of ChatGPT, a chatbot powered by OpenAI’s GPT-3 large language model (LLM), in November 2022 ushered generative AI into the public consciousness, and other companies like Google and Microsoft have been equally busy creating new opportunities to leverage the technology. In the meantime, these continuing advancements and applications of generative AI have raised important questions about how the technology will affect the labor market, how its use of training data implicates intellectual property rights, and what shape government regulation of this industry should take. Last week, a congressional hearing with key industry leaders suggested an openness to AI regulation—something that legislators have already considered to reign in some of the potential negative consequences of generative AI and AI more broadly. Considering these developments, scholars across the Center for Technology Innovation (CTI) weighed in around the halls on what the regulation of generative AI should look like.
NICOL
TURNER LEE (@DrTurnerLee)
Generative AI refers to machine learning algorithms that can create new content
like audio, code, images, text, simulations, or even videos. More recent focus
has been on its enablement of chatbots, including ChatGPT, Bard, Copilot,
and other more sophisticated tools that leverage LLMs to
perform a variety of functions, like gathering research for assignments,
compiling legal case files, automating repetitive clerical tasks, or improving
online search. While debates around regulation are focused on the potential
downsides to generative AI, including the quality of datasets, unethical
applications, racial or gender bias, workforce implications, and greater
erosion of democratic processes due to technological manipulation by bad
actors, the upsides include a dramatic spike in efficiency and productivity as
the technology improves and simplifies certain processes and decisions like
streamlining physician processing of
medical notes, or helping educators teach critical
thinking skills. There will be a lot to discuss around generative AI’s ultimate
value and consequence to society, and if Congress continues to operate at a
very slow pace to regulate emerging technologies and institute a federal
privacy standard, generative AI will become more technically advanced and
deeply embedded in society. But where Congress could garner a very quick win on
the regulatory front is to require consumer disclosures when AI-generated
content is in use and add labeling or some type of multi-stakeholder certification
process to encourage improved transparency and accountability for existing and
future use cases.
Once again, the European
Union is already leading the way on this. In its most recent AI Act,
the EU requires that AI-generated content be disclosed to consumers to prevent
copyright infringement, illegal content, and other malfeasance related to
end-user lack of understanding about these systems. As more chatbots mine,
analyze, and present content in accessible ways for users, findings are often
not attributable to any one or multiple sources, and despite some permissions
of content use granted under the fair use doctrine in
the U.S. that protects copyright-protected work, consumers are often left in
the dark around the generation and explanation of the process and results.
Congress should prioritize
consumer protection in future regulation, and work to create agile policies
that are futureproofed to adapt to emerging consumer and societal
harms—starting with immediate safeguards for users before they are left to,
once again, fend for themselves as subjects of highly digitized products and
services. The EU may honestly be onto something with the disclosure
requirement, and the U.S. could further contextualize its application vis-à-vis
existing models that do the same, including the labeling guidance
of the Food and Drug Administration (FDA) or what I have proposed in prior
research: an adaptation of the Energy
Star Rating system to AI. Bringing more transparency and accountability
to these systems must be central to any regulatory framework, and beginning
with smaller bites of a big apple might be a first stab for policymakers.
NIAM
YARAGHI (@niamyaraghi)
With the emergence of sophisticated artificial intelligence (AI) advancements,
including large language models (LLMs) like GPT-4, and LLM-powered applications
like ChatGPT, there is a pressing need to revisit healthcare privacy
protections. At their core, all AI innovations utilize sophisticated
statistical techniques to discern patterns within extensive datasets using
increasingly powerful yet cost-effective computational technologies. These
three components—big data, advanced statistical methods, and computing
resources—have not only become available recently but are also being
democratized and made readily accessible to everyone at a pace unprecedented in
previous technological innovations. This progression allows us to identify
patterns that were previously indiscernible, which creates opportunities for
important advances but also possible harms to patients.
Privacy regulations, most
notably HIPAA, were established to protect patient confidentiality, operating
under the assumption that de-identified data would remain anonymous. However,
given the advancements in AI technology, the current landscape has become
riskier. Now, it’s easier than ever to integrate various datasets from multiple
sources, increasing the likelihood of accurately identifying individual
patients.
Apart from the amplified risk
to privacy and security, novel AI technologies have also increased the value of
healthcare data due to the enriched potential for knowledge extraction.
Consequently, many data providers may become more hesitant to share medical
information with their competitors, further complicating healthcare data
interoperability.
Considering these heightened
privacy concerns and the increased value of healthcare data, it’s crucial to
introduce modern legislation to ensure that medical providers will continue
sharing their data while being shielded against the consequences of potential
privacy breaches likely to emerge from the widespread use of generative AI.
MARK
MACCARTHY (@Mark_MacCarthy)
In “The
Leopard,” Giuseppe Di Lampedusa’s famous novel of the Sicilian
aristocratic reaction to the unification of Italy in the 1860s, one of his
central characters says, “If we want things to stay as they are, things will
have to change.”
Something like this Sicilian
response might be happening in the tech industry’s embrace of
inevitable AI regulation. Three things are needed, however, if we do not want
things to stay as they are.
The first and most important
step is sufficient resources for agencies to enforce current law. Federal Trade
Commission Chair Lina Khan properly says AI
is not exempt from current consumer protection, discrimination, employment, and
competition law, but if regulatory agencies cannot hire technical staff and
bring AI cases in a time of budget austerity, current law will be a dead
letter.
Second, policymakers should
not be distracted by science fiction fantasies of AI programs developing
consciousness and achieving independent agency over humans, even if these
metaphysical abstractions are endorsed by
industry leaders. Not a dime of public money should be spent on these highly
speculative diversions when scammers and industry edge-riders are seeking to
use AI to break existing law.
Third, Congress should
consider adopting new identification, transparency, risk assessment, and
copyright protection requirements along the lines of the European Union’s
proposed AI
Act. The National Telecommunications and Information
Administration’s request
for comment on a proposed AI accountability framework and Sen.
Chuck Schumer’s (D-NY) recently-announced legislative
initiative to regulate AI might be moving in that direction.
TOM
WHEELER (@tewheels)
Both sides of the political aisle, as well as digital corporate chieftains, are
now talking about the need to regulate AI. A common theme is the need for a new
federal agency. To simply clone the model used for existing regulatory agencies
is not the answer, however. That model, developed for oversight of an
industrial economy, took advantage of slower paced innovation to micromanage
corporate activity. It is unsuitable for the velocity of the free-wheeling AI
era.
All regulations walk a
tightrope between protecting the public interest and promoting innovation and
investment. In the AI era, traversing this path means accepting that different
AI applications pose different risks and identifying a plan that pairs the
regulation with the risk while avoiding innovation-choking regulatory
micromanagement.
Such agility begins with
adopting the formula by which digital companies create technical standards
as the formula for developing behavioral standards: identify
the issue; assemble a standard-setting process involving the companies, civil
society, and the agency; then give final approval and enforcement authority to
the agency.
Industrialization was all
about replacing and/or augmenting the physical power of
humans. Artificial intelligence is about replacing and/or augmenting
humans’ cognitive powers. To confuse how the former was
regulated with what is needed for the latter would be to miss the opportunity
for regulation to be as innovative as the technology it oversees. We need
institutions for the digital era that address problems that already are
apparent to all.
Google and Microsoft are
general, unrestricted donors to the Brookings Institution. The findings,
interpretations, and conclusions posted in this piece are solely those of the
author and are not influenced by any donation.
Neuralink and the Brain’s Magical Future
April 20, 2017 By Tim Urban
Okay maybe that’s not exactly how it happened, and maybe those weren’t his exact words. But after learning about the new company Elon Musk was starting, I’ve come to realize that that’s exactly what he’s trying to do.
When I wrote about Tesla and SpaceX, I learned that you can only fully wrap your head around certain companies by zooming both way, way in and way, way out. In, on the technical challenges facing the engineers, out on the existential challenges facing our species. In on a snapshot of the world right now, out on the big story of how we got to this moment and what our far future could look like.
Not only is Elon’s new venture—Neuralink—the same type of deal, but six weeks after first learning about the company, I’m convinced that it somehow manages to eclipse Tesla and SpaceX in both the boldness of its engineering undertaking and the grandeur of its mission. The other two companies aim to redefine what future humans will do—Neuralink wants to redefine what future humans will be.
The mind-bending bigness of Neuralink’s mission, combined with the labyrinth of impossible complexity that is the human brain, made this the hardest set of concepts yet to fully wrap my head around—but it also made it the most exhilarating when, with enough time spent zoomed on both ends, it all finally clicked. I feel like I took a time machine to the future, and I’m here to tell you that it’s even weirder than we expect.
But before I can bring you in the time machine to show you what I found, we need to get in our zoom machine—because as I learned the hard way, Elon’s wizard hat plans cannot be properly understood until your head’s in the right place.
So wipe your brain clean of what it thinks it knows about itself and its future, put on soft clothes, and let’s jump into the vortex.
___________
Contents
Part 6: The Great Merger : http://waitbutwhy.com/2017/04/neuralink.html
Vanity Fair
April 2017 issue
story excerpts from interview with Ray Kurzweil:
1. | Spiraling capabilities of self-improving artificial intelligence
Google has gobbled up almost every robotics & machine learning company.It bought Deep•Mind for $650 million, and built the Google Brain team to work on artificial intelligence.
Google hired Geoffrey Hinton, PhD, pioneer in artificial neural networks. It also hired Ray Kurzweil, the futurist who predicted humans are 28 years away from “singularity” — the moment when spiraling capabilities of self-improving artificial super-intelligence will exceed human intelligence, and humans will merge with AI to create hybrid beings of the future.
2. | Computers are already doing many attributes of thinking
Trying to puzzle out AI, I went to meet Ray Kurzweil — author of book The Singularity Is Near, a Utopian vision of AI future. Kurzweil said computers are already “doing many attributes of thinking. Just a few years ago, AI couldn’t tell the difference between a dog and cat. Now it can. The list of things humans can do better than computers gets smaller and smaller. We create these tools to extend our long reach.”
3. | The promise & peril are deeply intertwined, the strategy is control the peril
Ray Kurzweil uses the word “we” when talking about super-intelligent future beings — compared to Elon Musk’s more ominous “they.”
Elon Musk said he was bewildered that Ray Kurzweil isn’t worried about AI hazards. “That’s not true. I’m the one who articulated the dangers,” Kurzweil said. “The promise and peril are deeply intertwined. There are strategies to control the peril, as there have been with bio-tech guidelines.”
Kurzweil said Musk’s bête noire could come true. He said our AI children “may be friendly and may not. If it’s not friendly, we may have to fight it by getting an AI on our side that’s even smarter.”
4. | 3 stages of the human response to new tech
Kurzweil said there are 3 stages of the human response to new tech:
1. Wow!
2. Uh oh!
3. What other choice do we have but to move forward?
2. Uh oh!
3. What other choice do we have but to move forward?
Ray Kurzweil predicts by the 2030s, we will be cyborgs. Nanorobots the size of blood cells will heal our bodies from the inside — connecting us to synthetic neo-cortex in the cloud, and to virtual & augmented reality. “We’ll be funnier, more musical, wiser,” he said.
Deep neural network models score higher
than humans in reading and comprehension test
Case | Man with quadriplegia employs injury bridging technologies to move again – just by thinking
‘Mind reading’ technology identifies complex thoughts, using machine
learning and fMRI
CMU aims to map all types of knowledge
in the brain
June 30, 2017
By combining
machine-learning algorithms with fMRI brain imaging technology, Carnegie
Mellon University (CMU) scientists have discovered, in essense, how to “read
minds.”
The researchers used
functional magnetic resonance imaging (fMRI) to view how the brain encodes
various thoughts (based on blood-flow patterns in the brain). They discovered
that the mind’s building blocks for constructing complex thoughts are formed,
not by words, but by specific combinations of the brain’s various sub-systems.
Following up on
previous research, the findings, published in Human Brain Mapping (open-access preprint here) and funded by the U.S. Intelligence
Advanced Research Projects Activity (IARPA), provide new evidence that the neural dimensions of concept
representation are universal across people and languages.
“One of the big
advances of the human brain was the ability to combine individual concepts into
complex thoughts, to think not just of ‘bananas,’ but ‘I like to eat bananas in
evening with my friends,’” said CMU’s Marcel Just, the D.O. Hebb University Professor
of Psychology in the Dietrich College of
Humanities and Social Sciences. “We have finally developed a way to see
thoughts of that complexity in the fMRI signal. The discovery of this
correspondence between thoughts and brain activation patterns tells us what the
thoughts are built of.”
Goal: A brain map of
all types of knowledge
The researchers used
240 specific events (described by sentences such as “The storm destroyed the
theater”) in the study, with seven adult participants. They measured the
brain’s coding of these events using 42 “neurally plausible semantic features”
— such as person, setting, size, social interaction, and physical action (as
shown in the word clouds in the illustration above). By measuring the specific
activation of each of these 42 features in a person’s brain system, the program
could tell what types of thoughts that person was focused on.
The researchers used a
computational model to assess how the detected brain activation patterns (shown
in the top illustration, for example) for 239 of the event sentences
corresponded to the detected neurally plausible semantic features that
characterized each sentence. The program was then able to decode the features
of the 240th left-out sentence. (For “cross-validation,” they did the same for
the other 239 sentences.)
The model was able to
predict the features of the left-out sentence with 87 percent accuracy, despite
never being exposed to its activation before. It was also able to work in the
other direction: to predict the activation pattern of a previously unseen
sentence, knowing only its semantic features.
“Our method overcomes
the unfortunate property of fMRI to smear together the signals emanating from
brain events that occur close together in time, like the reading of two
successive words in a sentence,” Just explained. “This advance makes it
possible for the first time to decode thoughts containing several concepts.
That’s what most human thoughts are composed of.”
“A next step might be
to decode the general type of topic a person is thinking about, such as geology
or skateboarding,” he added. “We are on the way to making a map of all the
types of knowledge in the brain.”
Future possibilities
It’s conceivable that
the CMU brain-mapping method might be combined one day with other “mind
reading” methods, such as UC Berkeley’s method for using fMRI and computational
models to decode and reconstruct
people’s imagined visual experiences. Plus whatever Neuralink discovers.
Or if the CMU method
could be replaced by noninvasive functional near-infrared spectroscopy
(fNIRS), Facebook’s Building8 research
concept (proposed by
former DARPA head Regina Dugan) might be incorporated (a filter for creating
quasi ballistic photons, avoiding diffusion and creating a narrow beam for
precise targeting of brain areas, combined with a new method of detecting blood-oxygen
levels).
Using fNIRS might also
allow for adapting the method to infer thoughts of locked-in paralyzed
patients, as in the Wyss Center for Bio and
Neuroengineering research. It might even lead to ways to generally enhance human communication.
The CMU research is
supported by the Office of the Director of National Intelligence (ODNI) via the
Intelligence Advanced Research Projects Activity (IARPA) and the Air Force
Research Laboratory (AFRL).
CMU has created some
of the first cognitive tutors, helped to develop the Jeopardy-winning Watson,
founded a groundbreaking doctoral program in neural computation, and is the
birthplace of artificial intelligence and cognitive psychology. CMU also
launched BrainHub, an initiative that focuses on how the
structure and activity of the brain give rise to complex behaviors.
Abstract of Predicting the Brain Activation Pattern Associated With
the Propositional Content of a Sentence: Modeling Neural Representations of
Events and States
Even though much has recently been learned about the neural representation
of individual concepts and categories, neuroimaging research is only beginning
to reveal how more complex thoughts, such as event and state descriptions, are
neurally represented. We present a predictive computational theory of the
neural representations of individual events and states as they are described in
240 sentences. Regression models were trained to determine the mapping between
42 neurally plausible semantic features (NPSFs) and thematic roles of the
concepts of a proposition and the fMRI activation patterns of various cortical
regions that process different types of information. Given a semantic
characterization of the content of a sentence that is new to the model, the
model can reliably predict the resulting neural signature, or, given an
observed neural signature of a new sentence, the model can predict its semantic
content. The models were also reliably generalizable across participants. This
computational model provides an account of the brain representation of a
complex yet fundamental unit of thought, namely, the conceptual content of a
proposition. In addition to characterizing a sentence representation at the
level of the semantic and thematic features of its component concepts, factor
analysis was used to develop a higher level characterization of a sentence,
specifying the general type of event representation that the sentence evokes
(e.g., a social interaction versus a change of physical state) and the voxel
locations most strongly associated with each of the factors.
by Neil Sahota October 1, 2018
Ray Kurzweil — futurist, inventor, and best selling
author — said: “Our technology, our machines, is part of our humanity. We
created them to extend ourselves, and that is what is unique about human
beings.”
In the past few years, there has been considerable
discussion that we’re slowly merging with our technology — that humans are
becoming trans-human with updated abilities: including
enhanced intelligence, strength, and awareness.
Considering Kurzweil’s words is a good place to begin
this talks. It’s no secret that Google has trans-humanistic aspirations. In
2011 author Steven Levy made this bold statement about Google in his book In
the Plex: “From the very start, Google’s founders saw the company as a
vehicle to realize the dream of artificial intelligence in augmenting humanity.”
It makes sense Google would bring on Ray Kurzweil to
be one of its Directors of Engineering in 2012. For years, Kurzweil has been
pushing the cultural conversation toward the idea of human transcendence with
his thought-provoking books.
Kurzweil has gained notoriety for the proposing
provocative idea: “The singularity will represent the culmination of the merger
of our biological thinking + existence with our technology — resulting in a
world that is still human but that transcends our biological roots.” But the
term “singularity” originated in a 1993 essay titled: The Coming
Technological Singularity — by science fiction author Vernor
Vinge PhD.
To grasp the significance of Vinge’s thinking: it’s
important to realize where we were as a society in the early 1990s. Back then,
smart-phones and social media websites were years away. The web — so vital to
all aspects of our life: communication, commerce, entertainment — was in its
infancy. But Vinge boldly proclaimed: “In 30 years we’ll have the technological
means to create super-human intelligence. Shortly after, the human era will be
ended.”
Here we are, almost 30 years from Vinge’s prediction
and the reality of trans-humanism has caught-on with the general public as a
distinct possibility. Writer Michael Ashley — co-author of my book Uber
Yourself Before You Get Kodaked: a modern primer on AI for the modern business
— and I sought to tap into the cultural zeitgeist on this topic, by
interviewing Ben Goertzel PhD.
Goertzel is the right person to speak about human
potential in the age of AI. He’s founder and CEO of the company SingularityNet.
Along with robotics engineer David Hanson of the company Hanson Robotics,
Goertzel co-created Sophia — the first robot to gain national citizenship.
Like Vinge and Kurzweil, Ben Goertzel PhD is
fascinated by the idea of trans-humanism. He explains it’s not pie-in-the-sky
conjecture — trans-humanism has been happening for awhile in analog form.
Goertzel said: “It’s happening bit-by-bit. If you take my glasses away, I’d become
heavily impaired and couldn’t participate in the world.”
He points to subtle ways humans are already merging
with computers. He said: “If you take the smart-phone away from my wife or
kids, they will go into withdrawal and also become heavily impaired.”
Still, many people fear trans-humanism. Critics warn
of designer babies and chips implanted in our minds. Theologians fear we will
denigrate the soul’s sanctity by achieving immortality. In the early 2000s, the
editors of Foreign Policy asked policy intellectuals: “What
idea, if embraced, would pose the greatest threat to the welfare of humanity?”
Francis Fukuyama, professor of international political economy at Johns Hopkins
School of Advanced International Studies, pointed to transhumanism, calling it
the “world’s most dangerous idea.” Writing for Psychology Today, Massimo Pigliucci stated,
“There are several problems with the pursuit of immortality, one of which is
particularly obvious. If we all live (much, much) longer, we all consume more
resources and have more children, leading to even more overpopulation and environmental
degradation.”
No matter the intellectual misgivings surrounding this
controversial topic, the fact remains that if we view transhumanism the way it
is conventionally defined, people have been evolving toward an updated version
of humanity for some time. “In some ways, we already operate as human
machine-hybrids,” said Goertzel. “If a caveman came into the modern world, he
would be astounded at how symbiotic we are with the various machines we use. We
use cars to get from point A to point B and air conditioners to regulate our
temperature. In Hong Kong at least, you never see anyone who’s not holding a
phone in their hand and staring at it.”
But there may be other, more pragmatic reasons why we
need to become transhuman, if only to stand up to the intelligent machines that
are coming. Early on, Elon Musk sounded the alarm about humans being usurped by
artificial intelligence in a series of well-publicized warnings. Since then, he has
suggested that the only way not to be overtaken by computers is to merge with
our creations. His venture, Neuralink, is in development precisely for this
purpose. Meant to combine human brains with computers, it’s his attempt to
symbiotically join our minds with the machines. “The merge scenario with A.I.
is the one that seems like probably the best,” he recently said on the podcast,
the Joe Rogan
Experience. “If you can’t beat it, join it.”
A visionary himself, Goertzel has long foreseen Musk’s
vision coming, yet he urges caution in its implementation. “The next step to
take is to wire these machines directly into the brain and body rather than
have them held in our hands. Clearly, this takes time and thought because you
need to be careful with sticking wires into human brains and bodies. But that
work is being done, and it’s not going to take more than a decade.”
Returning to Vinge’s prescience at the end of the 20th
century, we can see he was imagining a future that would occur even sooner than
he predicted. If we take Goertzel at his word, we are through Fukuyama’s and
others’ hand-wringing stage. We’re now at the point to think about
practicalities. Technology slows down for no one. Whether we like it or not,
there is a pre-smartphone and a post-smartphone world. Presumably, we all know
someone who was loathe to adopt the new technology — it’s likely their business
even suffered until they began using an iPhone or Android — or got swept aside
by adapters willing to change with the times. Are we at the precipice of a
similar phenomenon? Are we staring down the gulf at “Human 2.0?”
To put this dilemma in clearer focus, Goertzel advises
considering the question, not from your perspective, but from your child’s. He
paints a picture: “Imagine it’s eight years from now. All the other kids in
your daughter’s third-grade class are way ahead of her because their brains are
connected directly to Google and a calculator, and they’re SMSing back and
forth by Wi-Fi telepathy between their brains while your daughter sits there in
class being stunted because she must memorize things the old-fashioned way and
can’t send messages brain-to-brain.”
Goertzel suggests you consider what you would do if
your daughter’s teacher brought you in for a parent conference and told you
your daughter couldn’t keep up with her classmates. Imagine she suggested some form
of upgrade. You love your daughter. You want the best for her. What
would you do?
At this point, the prospect of trans-humanism stops
being an intellectual exercise. It becomes a question of subsistence.
An artificial synapse for future
miniaturized portable ‘brain-on-a-chip’ devices
MIT engineers
plan a fingernail-size chip that could replace a supercomputer
January 22, 2018
MIT engineers have designed
a new artificial synapse made from silicon germanium that can precisely control
the strength of an electric current flowing across it.
In simulations, the
researchers found that the chip and its synapses could be used to recognize
samples of handwriting with 95 percent accuracy. The engineers say the new
design, published today (Jan. 22) in the journal Nature Materials,
is a major step toward building portable, low-power neuromorphic chips for use
in pattern recognition and other machine-learning tasks.
Controlling the flow of ions: the
challenge
Researchers in the emerging
field of “neuromorphic computing” have attempted to design computer chips that
work like the human brain. The idea is to apply a voltage across layers that
would cause ions (electrically charged atoms) to move in a switching medium
(synapse-like space) to create conductive filaments in a manner that’s similar
to how the “weight” (connection strength) of a synapse changes.
There are more than 100
trillion synapses (in a typical human brain) that mediate neuron signaling in
the brain, strengthening some neural connections while pruning (weakening)
others — a process that enables the brain to recognize patterns, remember
facts, and carry out other learning tasks, all at lightning speeds.
Instead of carrying out
computations based on binary, on/off signaling, like current digital chips, the
elements of a “brain on a chip” would work in an analog fashion, exchanging a
gradient of signals, or “weights” — much like neurons that activate in various
ways (depending on the type and number of ions that flow across a synapse).
But it’s been difficult to
control the flow of ions in existing synapse designs. These have multiple paths
that make it difficult to predict where ions will make it through, according to
research team leader Jeehwan Kim, PhD, an assistant professor in the
departments of Mechanical Engineering and Materials Science and Engineering, a
principal investigator in MIT’s Research Laboratory of Electronics and
Microsystems Technology Laboratories.
“Once you apply some voltage
to represent some data with your artificial neuron, you have to erase and be
able to write it again in the exact same way,” Kim says. “But in an amorphous
solid, when you write again, the ions go in different directions because there
are lots of defects. This stream is changing, and it’s hard to control. That’s
the biggest problem — nonuniformity of the artificial synapse.”
So instead of using
amorphous materials as an artificial synapse, Kim and his colleagues created an
new “epitaxial random access memory” (epiRAM) design.
They started with a wafer of
silicon. They then grew a similar pattern of silicon germanium — a material
used commonly in transistors — on top of the silicon wafer. Silicon germanium’s
lattice is slightly larger than that of silicon, and Kim found that together,
the two perfectly mismatched materials could form a funnel-like dislocation,
creating a single path through which ions can predictably flow.*
This is the most uniform
device we could achieve, which is the key to demonstrating artificial neural
networks,” Kim says.
Testing the ability to recognize samples
of handwriting
As a test, Kim and his team
explored how the epiRAM device would perform if it were to carry out an actual
learning task: recognizing samples of handwriting — which researchers consider
to be a practical test for neuromorphic chips. Such chips would consist of
artificial “neurons” connected to other “neurons” via filament-based artificial
“synapses.”
They ran a computer
simulation of an artificial neural network consisting of three sheets of neural
layers connected via two layers of artificial synapses, based on measurements
from their actual neuromorphic chip. They fed into their simulation tens of
thousands of samples from the MNIST handwritten recognition dataset**, commonly used by neuromorphic designers.
They found that their neural
network device recognized handwritten samples 95.1 percent of the time — close
to the 97 percent accuracy of existing software algorithms running on large
computers.
A chip to replace a supercomputer
The team is now in the
process of fabricating a real working neuromorphic chip that can carry out
handwriting-recognition tasks. Looking beyond handwriting, Kim says the team’s
artificial synapse design will enable much smaller, portable neural network
devices that can perform complex computations that are currently only possible
with large supercomputers.
“Ultimately, we want a chip
as big as a fingernail to replace one big supercomputer,” Kim says. “This opens
a stepping stone to produce real artificial intelligence hardware.”
This research was supported
in part by the National Science Foundation. Co-authors included researchers at
Arizona State University.
* They applied
voltage to each synapse and found that all synapses exhibited about the same
current, or flow of ions, with about a 4 percent variation between synapses — a
much more uniform performance compared with synapses made from amorphous
material. They also tested a single synapse over multiple trials, applying the
same voltage over 700 cycles, and found the synapse exhibited the same current,
with just 1 percent variation from cycle to cycle.
** The MNIST
(Modified National Institute of Standards and Technology database) is a large database of handwritten
digits that is commonly used for training various image processing systems
and for training and testing in the field of machine learning. It contains 60,000 training images and
10,000 testing images.
Abstract of SiGe epitaxial memory
for neuromorphic computing with reproducible high performance based on
engineered dislocations
Although several types of
architecture combining memory cells and transistors have been used to
demonstrate artificial synaptic arrays, they usually present limited
scalability and high power consumption. Transistor-free analog switching
devices may overcome these limitations, yet the typical switching process they
rely on—formation of filaments in an amorphous medium—is not easily controlled
and hence hampers the spatial and temporal reproducibility of the performance.
Here, we demonstrate analog resistive switching devices that possess
desired characteristics for neuromorphic computing networks with minimal
performance variations using a single-crystalline SiGe layer epitaxially grown
on Si as a switching medium. Such epitaxial random access memories utilize
threading dislocations in SiGe to confine metal filaments in a defined,
one-dimensional channel. This confinement results in drastically enhanced
switching uniformity and long retention/high endurance with a high analog
on/off ratio. Simulations using the MNIST handwritten recognition data set
prove that epitaxial random access memories can operate with an online learning
accuracy of 95.1%.
references:
IBM researchers use analog memory to train
deep neural networks faster and more efficiently
New approach
allows deep neural networks to run hundreds of times faster than with GPUs,
using hundreds of times less energy
June 15, 2018
Crossbar arrays
of non-volatile memories can accelerate the training of neural networks by
performing computation at the actual location of the data. (credit: IBM
Research)
Imagine advanced artificial
intelligence (AI) running on your smartphone — instantly presenting the
information that’s relevant to you in real time. Or a supercomputer that
requires hundreds of times less energy.
The IBM Research AI team has
demonstrated a new approach that they believe is a major step toward those
scenarios.
Deep neural networks
normally require fast, powerful graphical processing unit (GPU) hardware
accelerators to support the needed high speed and computational accuracy —
such as the GPU devices used in the just-announced Summit supercomputer. But GPUs are highly energy-intensive,
making their use expensive and limiting their future growth, the researchers explain
in a recent paper published in Nature.
Analog memory replaces software,
overcoming the “von Neumann bottleneck”
Instead, the IBM researchers
used large arrays of non-volatile analog
memory devices (which use continuously variable signals rather than binary 0s
and 1s) to perform computations. Those arrays allowed the researchers to
create, in hardware, the same scale and precision of AI calculations that are
achieved by more energy-intensive systems in software, but running hundreds of
times faster and at hundreds of times lower power — without sacrificing the
ability to create deep learning systems.*
The trick was to replace
conventional von Neumann
architecture, which is “constrained by the time and energy spent moving
data back and forth between the memory and the processor (the ‘von Neumann
bottleneck’),” the researchers explain in the paper. “By contrast, in a non-von
Neumann scheme, computing is done at the location of the data [in memory], with
the strengths of the synaptic connections (the ‘weights’) stored and adjusted
directly in memory.
“Delivering the future of AI
will require vastly expanding the scale of AI calculations,” they note.
“Instead of shipping digital data on long journeys between digital memory chips
and processing chips, we can perform all of the computation inside the
analog memory chip. We believe this is a major step on the path to the kind of
hardware accelerators necessary for the next AI breakthroughs.”**
Given these encouraging
results, the IBM researchers have already started exploring the design of
prototype hardware accelerator chips, as part of an IBM Research
Frontiers Institute project, they said.
Ref.: Nature.
Source: IBM Research
* “From
these early design efforts, we were able to provide, as part of our Nature paper, initial estimates for the
potential of such [non-volatile memory]-based chips for training
fully-connected layers, in terms of the computational energy efficiency (28,065
GOP/sec//W) and throughput-per-area (3.6 TOP/sec/mm2). These values exceed
the specifications of today’s GPUs by two orders of
magnitude. Furthermore, fully-connected layers are a type of neural
network layer for which actual GPU performance frequently falls well below the
rated specifications. … Analog non-volatile memories can efficiently
accelerate at the heart of many recent AI advances. These memories allow the
“multiply-accumulate” operations used throughout these algorithms to be
parallelized in the analog domain, at the location of weight data, using
underlying physics. Instead of large circuits to multiply and add digital
numbers together, we simply pass a small current through a resistor into a wire,
and then connect many such wires together to let the currents build
up. This lets us perform many calculations at the same time, rather than
one after the other.”
** “By combining
long-term storage in phase-change memory (PCM) devices, near-linear update of
conventional complementary metal-oxide semiconductor (CMOS) capacitors and
novel techniques for cancelling out device-to-device variability, we finessed
these imperfections and achieved software-equivalent DNN accuracies on a
variety of different networks. These experiments used a mixed hardware-software
approach, combining software simulations of system elements that are easy to
model accurately (such as CMOS devices) together with full hardware
implementation of the PCM devices. It was essential to use real analog
memory devices for every weight in our neural networks, because modeling
approac
"Update
your AGI predictions"--- Prof. Roman Yampolskiy, PhD @romanyam
January 15, 2018
Microsoft and Alibaba have
developed deep neural network models that scored higher than humans in a
Stanford University reading and comprehension test, Stanford Question Answering Dataset (SQuAD).
Microsoft achieved 82.650 on
the ExactMatch (EM) metric* on Jan. 3, and Alibaba
Group Holding Ltd. scored 82.440 on Jan. 5. The best human score so
far is 82.304.
“SQuAD is a new reading
comprehension dataset, consisting of questions posed by crowdworkers on a set
of Wikipedia articles, where the answer to every question is a segment of text,
or span, from the corresponding reading passage,” according to the
Stanford NLP Group. “With 100,000+ question-answer pairs on 500+ articles,
SQuAD is significantly larger than previous reading comprehension datasets.”
“A strong start to 2018 with
the first model (SLQA+) to exceed human-level performance on @stanfordnlp SQuAD’s EM metric!,”
said Pranav Rajpurkar, a
Ph.D. student in the Stanford
Machine Learning Group and lead author of a paper in Proceedings
of the 2016 Conference on Empirical Methods in Natural Language Processing on
SQuAD (available on open-access ArXiv). “Next challenge: the F1 metric*, where humans still
lead by ~2.5 points!” (Alibaba’s SLQA+ scored 88.607 on the F1 metric and
Microsoft’s r-net+ scored 88.493.)
However, challenging the
“comprehension” description, Gary Marcus,
PhD, a Professor of Psychology and Neural Science at NYU, notes in a tweet
that “the SQUAD test shows that machines can highlight relevant passages in
text, not that they understand those passages.”
“The Chinese e-commerce
titan has joined the likes of Tencent
Holdings Ltd. and Baidu
Inc. in a race to develop AI that can enrich social media feeds,
target ads and services or even aid in autonomous driving, Bloomberg notes.
“Beijing has endorsed the technology in a national-level plan that calls for
the country to become the industry leader 2030.”
*”The ExactMatch
metric measures the percentage of predictions that match any one of the ground
truth answers exactly. The F1 score metric measures the average overlap between
the prediction and ground truth answer.” – Pranav Rajpurkar et al., ArXiv
Topics: AI/Robotics
Can we stop AI
outsmarting humanity?
The spectre of
superintelligent machines doing us harm is not just science fiction,
technologists say – so how can we ensure AI remains ‘friendly’ to its makers?
By Mara Hvistendahl
Thu 28 Mar 2019
It
began three and a half billion years ago in a pool of muck, when a molecule
made a copy of itself and so became the ultimate ancestor of all earthly life.
It began four million years ago, when brain volumes began climbing rapidly in
the hominid line.
Fifty thousand years ago with the rise of
Homo sapiens sapiens.
Ten thousand years ago with the invention
of civilization.
Five hundred years ago with the invention
of the printing press.
Jaan Tallinn stumbled
across these words in 2007, in an online essay called Staring
into the Singularity. The “it” was human civilisation. Humanity
would cease to exist, predicted the essay’s author, with the emergence of
superintelligence, or AI, that surpasses human-level intelligence in a broad
array of areas.
Tallinn, an Estonia-born
computer programmer, has a background in physics and a propensity to approach
life like one big programming problem. In 2003, he co-founded Skype, developing
the backend for the app. He cashed in his shares after eBay bought it two years
later, and now he was casting about for something to do. Staring into the
Singularity mashed up computer code, quantum physics and Calvin and Hobbes
quotes. He was hooked.
Tallinn soon discovered
that the author, Eliezer Yudkowsky, a self-taught theorist, had written more
than 1,000 essays and blogposts, many of them devoted to superintelligence. He
wrote a program to scrape Yudkowsky’s writings from the internet, order them
chronologically and format them for his iPhone. Then he spent the better part
of a year reading them.
The term artificial
intelligence, or the simulation of intelligence in computers or machines, was
coined back in 1956, only a decade after the creation of the first electronic
digital computers. Hope for the field was initially high, but by the 1970s,
when early predictions did not pan out, an “AI winter” set in. When Tallinn
found Yudkowsky’s essays, AI was undergoing a renaissance. Scientists were
developing AIs that excelled in specific areas, such as winning at chess,
cleaning the kitchen floor and recognising human speech. Such “narrow” AIs, as
they are called, have superhuman capabilities, but only in their specific areas
of dominance. A chess-playing AI cannot clean the floor or take you from point
A to point B. Superintelligent AI, Tallinn came to believe, will combine a wide
range of skills in one entity. More darkly, it might also use data generated by
smartphone-toting humans to excel at social manipulation.
Reading Yudkowsky’s
articles, Tallinn became convinced that superintelligence could lead to an
explosion or breakout of AI that could threaten human existence – that
ultrasmart AIs will take our place on the evolutionary ladder and dominate us
the way we now dominate apes. Or, worse yet, exterminate us.
After finishing the last of
the essays, Tallinn shot off an email to Yudkowsky – all lowercase, as is his
style. “i’m jaan, one of the founding engineers of skype,” he wrote. Eventually
he got to the point: “i do agree that ... preparing for the event of general AI
surpassing human intelligence is one of the top tasks for humanity.” He wanted
to help.
When Tallinn flew to the
Bay Area for other meetings a week later, he met Yudkowsky, who lived nearby,
at a cafe in Millbrae, California. Their get-together stretched to four hours.
“He actually, genuinely understood the underlying concepts and the details,”
Yudkowsky told me recently. “This is very rare.” Afterward, Tallinn wrote a
check for $5,000 (£3,700) to the Singularity Institute for Artificial
Intelligence, the nonprofit where Yudkowsky was a research fellow. (The
organisation changed its name to Machine Intelligence Research Institute,
or Miri, in 2013.) Tallinn has since given the institute more than $600,000.
The encounter with
Yudkowsky brought Tallinn purpose, sending him on a mission to save us from our
own creations. He embarked on a life of travel, giving talks around the world
on the threat posed by superintelligence. Mostly, though, he began funding
research into methods that might give humanity a way out: so-called friendly AI.
That doesn’t mean a machine or agent is particularly skilled at chatting about
the weather, or that it remembers the names of your kids – although
superintelligent AI might be able to do both of those things. It doesn’t mean
it is motivated by altruism or love. A common fallacy is assuming that AI has
human urges and values. “Friendly” means something much more fundamental: that
the machines of tomorrow will not wipe us out in their quest to attain their
goals.
Last spring, I joined Tallinn for a meal in the
dining hall of Cambridge University’s Jesus College. The churchlike space is
bedecked with stained-glass windows, gold moulding, and oil paintings of men in
wigs. Tallinn sat at a heavy mahogany table, wearing the casual garb of Silicon
Valley: black jeans, T-shirt and canvas sneakers. A vaulted timber ceiling
extended high above his shock of grey-blond hair.
At 47, Tallinn is in some
ways your textbook tech entrepreneur. He thinks that thanks to advances in
science (and provided AI doesn’t destroy us), he will live for “many, many
years”. When out clubbing with researchers, he outlasts even the young graduate
students. His concern about superintelligence is common among his cohort.
PayPal co-founder Peter
Thiel’s foundation has given $1.6m to Miri and, in 2015, Tesla
founder Elon
Musk donated $10m to the Future of Life Institute, a technology
safety organisation in Cambridge, Massachusetts. But Tallinn’s entrance to this
rarefied world came behind the iron curtain in the 1980s, when a classmate’s father
with a government job gave a few bright kids access to mainframe computers.
After Estonia became independent, he founded a video-game company. Today,
Tallinn still lives in its capital city – also called Tallinn – with his wife
and the youngest of his six kids. When he wants to meet with researchers, he
often just flies them to the Baltic region.
His giving strategy is
methodical, like almost everything else he does. He spreads his money among 11
organisations, each working on different approaches to AI safety, in the hope
that one might stick. In 2012, he cofounded the Cambridge Centre for the Study of
Existential Risk (CSER) with an initial outlay of close to
$200,000.
Existential risks – or
X-risks, as Tallinn calls them – are threats to humanity’s survival. In
addition to AI, the 20-odd researchers at CSER study climate change, nuclear
war and bioweapons. But, to Tallinn, those other disciplines “are really just
gateway drugs”. Concern about more widely accepted threats, such as climate
change, might draw people in. The horror of superintelligent machines taking
over the world, he hopes, will convince them to stay. He was visiting Cambridge
for a conference because he wants the academic community to take AI safety more
seriously.
At Jesus College, our
dining companions were a random assortment of conference-goers, including a
woman from Hong Kong who was studying robotics and a British man who graduated
from Cambridge in the 1960s. The older man asked everybody at the table where
they attended university. (Tallinn’s answer, Estonia’s University of Tartu, did
not impress him.) He then tried to steer the conversation toward the news.
Tallinn looked at him blankly. “I am not interested in near-term risks,” he
said.
Tallinn changed the topic
to the threat of superintelligence. When not talking to other programmers, he
defaults to metaphors, and he ran through his suite of them: advanced AI can
dispose of us as swiftly as humans chop down trees. Superintelligence is to us
what we are to gorillas.
An AI would need a body to
take over, the older man said. Without some kind of physical casing, how could
it possibly gain physical control?
Tallinn had another
metaphor ready: “Put me in a basement with an internet connection, and I could
do a lot of damage,” he said. Then he took a bite of risotto.
Every AI, whether it’s a Roomba or one of its
potential world-dominating descendants, is driven by outcomes. Programmers
assign these goals, along with a series of rules on how to pursue them.
Advanced AI wouldn’t necessarily need to be given the goal of world domination
in order to achieve it – it could just be accidental. And the history of
computer programming is rife with small errors that sparked catastrophes. In
2010, for example, when a trader with the mutual-fund company Waddell &
Reed sold thousands of futures contracts, the firm’s software left out a key
variable from the algorithm that helped execute the trade. The result was the
trillion-dollar US “flash
crash”.
The researchers Tallinn
funds believe that if the reward structure of a superhuman AI is not properly
programmed, even benign objectives could have insidious ends. One well-known
example, laid out by the Oxford University philosopher Nick
Bostrom in his book Superintelligence,
is a fictional agent directed to make as many paperclips as possible. The AI
might decide that the atoms in human bodies would be better put to use as raw
material.
A man plays chess
with a robot designed by Taiwan’s Industrial Technology Research Institute
(ITRI) in Taipei in 2017. Photograph: Sam Yeh/AFP/Getty
Tallinn’s views have their
share of detractors, even among the community of people concerned with AI
safety. Some object that it is too early to worry about restricting
superintelligent AI when we don’t yet understand it. Others say that focusing
on rogue technological actors diverts attention from the most urgent problems
facing the field, like the fact that the majority of algorithms are designed by
white men, or based on data biased
toward them. “We’re in danger of building a world that we don’t want
to live in if we don’t address those challenges in the near term,” said Terah
Lyons, executive director of the Partnership on AI, a technology industry
consortium focused on AI safety and other issues. (Several of the institutes
Tallinn backs are members.) But, she added, some of the near-term challenges
facing researchers, such as weeding out algorithmic bias, are precursors to
ones that humanity might see with super-intelligent AI.
Tallinn isn’t so convinced.
He counters that superintelligent AI brings unique threats. Ultimately, he
hopes that the AI community might follow the lead of the anti-nuclear movement
in the 1940s. In the wake of the bombings of Hiroshima and Nagasaki, scientists
banded together to try to limit further nuclear testing. “The Manhattan Project
scientists could have said: ‘Look, we are doing innovation here, and innovation
is always good, so let’s just plunge ahead,’” he told me. “But they were more
responsible than that.”
Tallinn warns that any approach to AI safety will be
hard to get right. If an AI is sufficiently smart, it might have a better
understanding of the constraints than its creators do. Imagine, he said,
“waking up in a prison built by a bunch of blind five-year-olds.” That is what
it might be like for a super-intelligent AI that is confined by humans.
The theorist Yudkowsky
found evidence this might be true when, starting in 2002, he conducted chat
sessions in which he played the role of an AI enclosed in a box, while a
rotation of other people played the gatekeeper tasked with keeping the AI in.
Three out of five times, Yudkowsky – a mere mortal – says he convinced the
gatekeeper to release him. His experiments have not discouraged researchers
from trying to design a better box, however.
The researchers that
Tallinn funds are pursuing a broad variety of strategies, from the practical to
the seemingly far-fetched. Some theorise about boxing AI, either physically, by
building an actual structure to contain it, or by programming in limits to what
it can do. Others are trying to teach AI to adhere to human values. A few are
working on a last-ditch off-switch. One researcher who is delving into all
three is mathematician and philosopher Stuart Armstrong at Oxford
University’s Future
of Humanity Institute, which Tallinn calls “the most interesting
place in the universe.” (Tallinn has given FHI more than $310,000.)
Armstrong is one of the few
researchers in the world who focuses full-time on AI safety. When I met him for
coffee in Oxford, he wore an unbuttoned rugby shirt and had the look of someone
who spends his life behind a screen, with a pale face framed by a mess of sandy
hair. He peppered his explanations with a disorienting mixture of
popular-culture references and math. When I asked him what it might look like
to succeed at AI safety, he said: “Have you seen the
Lego
movie? Everything is awesome.”
One strain of Armstrong’s
research looks at a specific approach to boxing called an “oracle” AI. In a
2012 paper with Nick Bostrom, who co-founded FHI, he proposed not only walling off
superintelligence in a holding tank – a physical structure – but also
restricting it to answering questions, like a really smart Ouija
board. Even with these boundaries, an AI would have immense power to
reshape the fate of humanity by subtly manipulating its interrogators. To
reduce the possibility of this happening, Armstrong proposes time limits on
conversations, or banning questions that might upend the current world order.
He also has suggested giving the oracle proxy measures of human survival, like
the Dow Jones industrial average or the number of people crossing the street in
Tokyo, and telling it to keep these steady.
Ultimately, Armstrong
believes, it could be necessary to create, as he calls it in one paper, a “big
red off button”: either a physical switch, or a mechanism programmed into an AI
to automatically turn itself off in the event of a breakout. But designing such
a switch is far from easy. It is not just that an advanced AI interested in
self-preservation could prevent the button from being pressed. It could also
become curious about why humans devised the button, activate it to see what
happens, and render itself useless. In 2013, a programmer named Tom Murphy VII
designed an AI that could teach itself to
play Nintendo Entertainment System games. Determined not to lose at Tetris, the
AI simply pressed pause – and kept the game frozen. “Truly, the only winning
move is not to play,” Murphy observed wryly, in a paper on his creation.
For the strategy to
succeed, an AI has to be uninterested in the button, or, as Tallinn put it: “It
has to assign equal value to the world where it’s not existing and the world
where it’s existing.” But even if researchers can achieve that, there are other
challenges. What if the AI has copied itself several thousand times across the
internet?
The approach that most
excites researchers is finding a way to make AI adhere to human values– not by
programming them in, but by teaching AIs to learn them. In a world dominated by
partisan politics, people often dwell on the ways in which our principles
differ. But, Tallinn told me, humans have a lot in common: “Almost everyone
values their right leg. We just don’t think about it.” The hope is that an AI
might be taught to discern such immutable rules.
In the process, an AI would
need to learn and appreciate humans’ less-than-logical side: that we often say
one thing and mean another, that some of our preferences conflict with others,
and that people are less reliable when drunk. Despite the challenges, Tallinn
believes, it is worth trying because the stakes are so high. “We have to think
a few steps ahead,” he said. “Creating an AI that doesn’t share our interests
would be a horrible mistake.”
On his last night in Cambridge, I joined Tallinn and
two researchers for dinner at a steakhouse. A waiter seated our group in a
white-washed cellar with a cave-like atmosphere. He handed us a one-page menu
that offered three different kinds of mash. A couple sat down at the table next
to us, and then a few minutes later asked to move elsewhere. “It’s too
claustrophobic,” the woman complained. I thought of Tallinn’s comment about the
damage he could wreak if locked in a basement with nothing but an internet
connection. Here we were, in the box. As if on cue, the men contemplated ways
to get out.
Tallinn’s guests included
former genomics researcher Seán Ó
hÉigeartaigh, who is CSER’s executive director, and Matthijs Maas,
an AI researcher at the University of Copenhagen. They joked about an idea for
a nerdy action flick titled Superintelligence v Blockchain!, and discussed an
online game called Universal Paperclips,
which riffs on the scenario in Bostrom’s book. The exercise involves repeatedly
clicking your mouse to make paperclips. It’s not exactly flashy, but it does
give a sense for why a machine might look for more expedient ways to produce
office supplies.
Eventually, talk shifted
toward bigger questions, as it often does when Tallinn is present. The ultimate
goal of AI-safety research is to create machines that are, as Cambridge
philosopher and CSER co-founder Huw Price once put it, “ethically as well as
cognitively superhuman”. Others have raised the question: if we don’t want AI
to dominate us, do we want to dominate AI? In other words, does AI have rights?
Tallinn believes this is needless anthropomorphising. It assumes that
intelligence equals consciousness – a misconception that annoys many AI
researchers. Earlier in the day, CSER researcher José Hernández-Orallo joked
that when speaking with AI researchers, consciousness is “the C-word”. (“And
‘free will’ is the F-word,” he added.)
In the cellar, Tallinn said
that consciousness is beside the point: “Take the example of a thermostat. No
one would say it is conscious. But it’s really inconvenient to face up against
that agent if you’re in a room that is set to negative 30 degrees.”
Ó hÉigeartaigh chimed in.
“It would be nice to worry about consciousness,” he said, “but we won’t have
the luxury to worry about consciousness if we haven’t first solved the
technical safety challenges.”
People get overly
preoccupied with what superintelligent AI is, Tallinn said. What form will it
take? Should we worry about a single AI taking over, or an army of them? “From
our perspective, the important thing is what AI does,” he stressed. And that,
he believes, may still be up to humans – for now.
This piece originally appeared in Popular Science magazine
AI will upload and access our memories, predicts Siri co-inventor
"Instead of asking how smart we can make our machines, let's ask how smart our machines can make us."
April 26, 2017
Instead of replacing humans with robots, artificial intelligence should be used more for augmenting human memory and other human weaknesses, Apple Inc. executive Tom Gruber suggested at the TED 2017 conference yesterday (April 25, 2017).
Thanks to the internet and our smartphones, much of our personal data is already being captured, notes Gruber, who was one the inventors of voice-controlled intelligent-assistant Siri. Future AI memory enhancement could be especially life-changing for those with Alzheimer’s or dementia, he suggested.
Limitless
“Superintelligence should give us super-human abilities,” he said. “As machines get smarter, so do we. Artificial intelligence can enable partnerships where each human on the team is doing what they do best. Instead of asking how smart we can make our machines, let’s ask how smart our machines can make us.
“I can’t say when or what form factors are involved, but I think it is inevitable,” he said. “What if you could have a memory that was as good as computer memory and is about your life? What if you could remember every person you ever met? How to pronounce their name? Their family details? Their favorite sports? The last conversation you had with them?”
Gruber’s ideas mesh with a prediction by Ray Kurzweil: “Once we have achieved complete models of human intelligence, machines will be capable of combining the flexible, subtle human levels of pattern recognition with the natural advantages of machine intelligence, in speed, memory capacity, and, most importantly, the ability to quickly share knowledge and skills.”
Two projects announced last week aim in that direction: Facebook’s plan to develop a non-invasive brain-computer interface that will let you type at 100 words per minute and Elon Musks’ proposal that we become superhuman cyborgs to deal with superintelligent AI.
But trusting machines also raises security concerns, Gruber warned. “We get to choose what is and is not recalled,” he said. “It’s absolutely essential that this be kept very secure.”
Quadriplegia patient uses brain-computer interface to move his arm by just
thinking
New Braingate design replaces robot arm
with muscle-stimulating system
April 26, 2017
Bill Kochevar, who was
paralyzed below his shoulders in a bicycling accident eight years ago, is the
first person with quadriplegia to have arm and hand movements restored without
robot help (credit: Case Western Reserve University/Cleveland FES Center)
A research team led by Case Western
Reserve University has developed the first implanted
brain-recording and muscle-stimulating system to restore arm and hand movements
for quadriplegic patients.*
In a proof-of-concept experiment, the system included a brain-computer
interface with recording electrodes implanted under his skull and a functional
electrical stimulation (FES) system that activated his arm and hand —
reconnecting his brain to paralyzed muscles.
The research was part of the ongoing BrainGate2 pilot
clinical trial being conducted by a consortium of academic and other
institutions to assess the safety and feasibility of the implanted
brain-computer interface (BCI) system in people with paralysis. Previous
Braingate designs required a robot arm.
In 2012 research, Jan
Scheuermann, who has quadriplegia, was able to feed herself using a
brain-machine interface and a computer-driven robot arm (credit: UPMC)
Kochevar’s eight years of muscle atrophy first required rehabilitation. The
researchers exercised Kochevar’s arm and hand with cyclical electrical
stimulation patterns. Over 45 weeks, his strength, range of motion. and
endurance improved. As he practiced movements, the researchers adjusted
stimulation patterns to further his abilities.
To prepare him to use his arm again, Kochevar learned how to use his own
brain signals to move a virtual-reality arm on a computer screen. The team then
implanted the FES systems’ 36 electrodes that animate muscles in the upper and
lower arm, allowing him to move the actual arm.
Kochevar can now make each joint in his right arm move individually. Or,
just by thinking about a task such as feeding himself or getting a drink, the
muscles are activated in a coordinated fashion.
Neural activity
(generated when Kochevar imagines movement of his arm and hand) is recorded
from two 96-channel microelectrode arrays implanted in the motor cortex, on the
surface of the brain. The implanted brain-computer interface translates the
recorded brain signals into specific command signals that determine the amount
of stimulation to be applied to each functional electrical stimulation (FES)
electrode in the hand, wrist, arm, elbow and shoulder, and to a mobile arm
support. (credit: A Bolu Ajiboye et al./The Lancet)
“Our research is at an early stage, but we believe that this
neuro-prosthesis could offer individuals with paralysis the possibility of
regaining arm and hand functions to perform day-to-day activities, offering
them greater independence,” said lead author Dr Bolu Ajiboye, Case Western
Reserve University. “So far, it has helped a man with tetraplegia to reach and
grasp, meaning he could feed himself and drink. With further development, we
believe the technology could give more accurate control, allowing a wider range
of actions, which could begin to transform the lives of people living with
paralysis.”
Work is underway to make the brain implant wireless, and the investigators
are improving decoding and stimulation patterns needed to make movements more
precise. Fully implantable FES systems have already been developed and are also
being tested in separate clinical research.
A study of the work was published in the The Lancet March
28, 2017.
Writing in a linked Comment to The Lancet, Steve Perlmutter,
M.D., University of Washington, said: “The goal is futuristic: a paralysed
individual thinks about moving her arm as if her brain and muscles were not
disconnected, and implanted technology seamlessly executes the desired
movement… This study is groundbreaking as the first report of a person
executing functional, multi-joint movements of a paralysed limb with a motor
neuro-prosthesis. However, this treatment is not nearly ready for use outside
the lab. The movements were rough and slow and required continuous visual
feedback, as is the case for most available brain-machine interfaces, and had
restricted range due to the use of a motorised device to assist shoulder
movements… Thus, the study is a proof-of-principle demonstration of what is
possible, rather than a fundamental advance in neuro-prosthetic concepts or
technology. But it is an exciting demonstration nonetheless, and the future of
motor neuro-prosthetics to overcome paralysis is brighter.”
* The study was funded by the US National Institutes of Health and the US
Department of Veterans Affairs. It was conducted by scientists from Case
Western Reserve University, Department of Veterans Affairs Medical Center,
University Hospitals Cleveland Medical Center, MetroHealth Medical Center,
Brown University, Massachusetts General Hospital, Harvard Medical School, Wyss
Center for Bio and Neuroengineering. The investigational BrainGate technology
was initially developed in the Brown University laboratory of John Donoghue,
now the founding director of the Wyss Center for Bio and Neuroengineering in
Geneva, Switzerland. The implanted recording electrodes are known as the Utah
array, originally designed by Richard Normann, Emeritus Distinguished Professor
of Bioengineering at the University of Utah. The report in Lancet is the result
of a long-running collaboration between Kirsch, Ajiboye and the
multi-institutional BrainGate consortium. Leigh Hochberg, a neurologist and
neuroengineer at Massachusetts General Hospital, Brown University and the VA
RR&D Center for Neurorestoration and Neurotechnology in Providence, Rhode
Island, directs the pilot clinical trial of the BrainGate system and is a study
co-author.
Case | Man with quadriplegia employs injury bridging technologies to move again – just by thinking
Abstract of Restoration of reaching and grasping movements through
brain-controlled muscle stimulation in a person with tetraplegia: a
proof-of-concept demonstration
Background: People with chronic tetraplegia, due to high-cervical spinal cord
injury, can regain limb movements through coordinated electrical stimulation of
peripheral muscles and nerves, known as functional electrical stimulation
(FES). Users typically command FES systems through other preserved, but
unrelated and limited in number, volitional movements (eg, facial muscle
activity, head movements, shoulder shrugs). We report the findings of an
individual with traumatic high-cervical spinal cord injury who coordinated
reaching and grasping movements using his own paralysed arm and hand,
reanimated through implanted FES, and commanded using his own cortical signals
through an intracortical brain–computer interface (iBCI).
Methods: We recruited a participant into the BrainGate2 clinical trial, an
ongoing study that obtains safety information regarding an intracortical neural
interface device, and investigates the feasibility of people with tetraplegia
controlling assistive devices using their cortical signals. Surgical procedures
were performed at University Hospitals Cleveland Medical Center (Cleveland, OH,
USA). Study procedures and data analyses were performed at Case Western Reserve
University (Cleveland, OH, USA) and the US Department of Veterans Affairs,
Louis Stokes Cleveland Veterans Affairs Medical Center (Cleveland, OH, USA).
The study participant was a 53-year-old man with a spinal cord injury (cervical
level 4, American Spinal Injury Association Impairment Scale category A). He
received two intracortical microelectrode arrays in the hand area of his motor
cortex, and 4 months and 9 months later received a total of 36 implanted
percutaneous electrodes in his right upper and lower arm to electrically
stimulate his hand, elbow, and shoulder muscles. The participant used a
motorised mobile arm support for gravitational assistance and to provide
humeral abduction and adduction under cortical control. We assessed the
participant’s ability to cortically command his paralysed arm to perform simple
single-joint arm and hand movements and functionally meaningful multi-joint
movements. We compared iBCI control of his paralysed arm with that of a virtual
three-dimensional arm. This study is registered with ClinicalTrials.gov,
number NCT00912041.
Findings: The intracortical implant occurred on Dec 1, 2014, and we are
continuing to study the participant. The last session included in this report
was Nov 7, 2016. The point-to-point target acquisition sessions began on Oct 8,
2015 (311 days after implant). The participant successfully cortically
commanded single-joint and coordinated multi-joint arm movements for
point-to-point target acquisitions (80–100% accuracy), using first a virtual
arm and second his own arm animated by FES. Using his paralysed arm, the
participant volitionally performed self-paced reaches to drink a mug of coffee
(successfully completing 11 of 12 attempts within a single session 463 days
after implant) and feed himself (717 days after implant).
Interpretation: To our knowledge, this is the first report of a combined implanted
FES+iBCI neuroprosthesis for restoring both reaching and grasping movements to
people with chronic tetraplegia due to spinal cord injury, and represents a
major advance, with a clear translational path, for clinically viable
neuroprostheses for restoration of reaching and grasping after paralysis.
Funding: National Institutes of Health, Department of Veterans Affairs.
references:
Topics: Cognitive Science/Neuroscience | ElectronicsTopics: AI/RoboticsWhat if you could type directly from your brain at 100 words per minute?
Former DARPA director reveals Facebook's secret research projects to create a non-invasive brain-computer interface and haptic skin hearing
April 19, 2017
Regina Dugan, PhD, Facebook VP of Engineering, Building8, revealed today (April 19, 2017) at Facebook F8 conference 2017 a plan to develop a non-invasive brain-computer interface that will let you type at 100 wpm — by decoding neural activity devoted to speech.
Dugan previously headed Google’s Advanced Technology and Projects Group, and before that, was Director of the Defense Advanced Research Projects Agency (DARPA).
She explained in a Facebook post that over the next two years, her team will be building systems that demonstrate “a non-invasive system that could one day become a speech prosthetic for people with communication disorders or a new means for input to AR [augmented reality].”
Dugan said that “even something as simple as a ‘yes/no’ brain click … would be transformative.” That simple level has been achieved by using functional near-infrared spectroscopy (fNIRS) to measure changes in blood oxygen levels in the frontal lobes of the brain, as KurzweilAI recently reported. (Near-infrared light can penetrate the skull and partially into the brain.)
Dugan agrees that optical imaging is the best place to start, but her Building8 team team plans to go way beyond that research — sampling hundreds of times per second and precise to millimeters. The research team began working on the brain-typing project six months ago and she now has a team of more than 60 researchers who specialize in optical neural imaging systems that push the limits of spatial resolution and machine-learning methods for decoding speech and language.
The research is headed by Mark Chevillet, previously an adjunct professor of neuroscience at Johns Hopkins University.
Besides replacing smartphones, the system would be a powerful speech prosthetic, she noted — allowing paralyzed patients to “speak” at normal speed.
Dugan revealed one specific method the researchers are currently working on to achieve that: a ballistic filter for creating quasi ballistic photons (avoiding diffusion) — creating a narrow beam for precise targeting — combined with a new method of detecting blood-oxygen levels.
Dugan also described a system that may one day allow hearing-impaired people to hear directly via vibrotactile sensors embedded in the skin. “In the 19th century, Braille taught us that we could interpret small bumps on a surface as language,” she said. “Since then, many techniques have emerged that illustrate our brain’s ability to reconstruct language from components.” Today, she demonstrated “an artificial cochlea of sorts and the beginnings of a new a ‘haptic vocabulary’.”
In a neurotechnology future, human-rights laws will need to be revisited
April 28, 2017
New human rights laws to prepare for rapid current advances in neurotechnology that may put “freedom of mind” at risk have been proposed in the open access journal Life Sciences, Society and Policy.
Four new human rights laws could emerge in the near future to protect against exploitation and loss of privacy, the authors of the study suggest: The right to cognitive liberty, the right to mental privacy, the right to mental integrity, and the right to psychological continuity.
Advances in neural engineering, brain imaging, and neurotechnology put freedom of the mind at risk, says Marcello Ienca, lead author and PhD student at the Institute for Biomedical Ethics at the University of Basel. “Our proposed laws would give people the right to refuse coercive and invasive neurotechnology, protect the privacy of data collected by neurotechnology, and protect the physical and psychological aspects of the mind from damage by the misuse of neurotechnology.”
Potential misuses
Sophisticated brain imaging and the development of brain-computer interfaces have moved away from a clinical setting into the consumer domain. There’s a risk that the technology could be misused and create unprecedented threats to personal freedom. For example:
- Uses in criminal court as a tool for assessing criminal responsibility or even the risk of re-offending.*
- Consumer companies using brain imaging for “neuromarketing” to understand consumer behavior and elicit desired responses from customers.
- “Brain decoders” that can turn a person’s brain imaging data into images, text or sound.**
- Hacking, allowing a third-party to eavesdrop on someone’s mind.***
International human rights laws currently make no specific mention of neuroscience. But as with the genetic revolution, the on-going neurorevolution will require consideration of human-rights laws and even the creation of new ones, the authors suggest.
* “A possibly game-changing use of neurotechnology in the legal field has been illustrated by Aharoni et al. (2013). In this study, researchers followed a group of 96 male prisoners at prison release. Using fMRI, prisoners’ brains were scanned during the performance of computer tasks in which they had to make quick decisions and inhibit impulsive reactions. The researchers followed the ex-convicts for 4 years to see how they behaved. The study results indicate that those individuals showing low activity in a brain region associated with decision-making and action (the Anterior Cingulate Cortex, ACC) are more likely to commit crimes again within 4 years of release (Aharoni et al. 2013). According to the study, the risk of recidivism is more than double in individuals showing low activity in that region of the brain than in individuals with high activity in that region. Their results suggest a “potential neurocognitive biomarker for persistent antisocial behavior”. In other words, brain scans can theoretically help determine whether certain convicted persons are at an increased risk of reoffending if released.” — Marcello Ienca and Roberto Andorno/Life Sciences, Society and Policy
** NASA and Jaguar are jointly developing a technology called Mind Sense, which will measure brainwaves to monitor the driver’s concentration in the car (Biondi and Skrypchuk 2017). If brain activity indicates poor concentration, then the steering wheel or pedals could vibrate to raise the driver’s awareness of the danger. This technology can contribute to reduce the number of accidents caused by drivers who are stressed or distracted. However, it also opens theoretically the possibility for third parties to use brain decoders to eavesdropping on people’s states of mind. — Marcello Ienca and Roberto Andorno/Life Sciences, Society and Policy
*** Criminally motivated actors could selectively erase memories from their victims’ brains to prevent being identified by them later on or simply to cause them harm. On the long term-scenario, they could be used by surveillance and security agencies with the purpose of selectively erasing dangerous, inconvenient from people’s brain as portrayed in the movie Men in Black with the so-called neuralyzer. — Marcello Ienca and Roberto Andorno/Life Sciences, Society and Policy
Abstract of Towards new human rights in the age of neuroscience and neurotechnology
Rapid advancements in human neuroscience and neurotechnology open unprecedented possibilities for accessing, collecting, sharing and manipulating information from the human brain. Such applications raise important challenges to human rights principles that need to be addressed to prevent unintended consequences. This paper assesses the implications of emerging neurotechnology applications in the context of the human rights framework and suggests that existing human rights may not be sufficient to respond to these emerging issues. After analysing the relationship between neuroscience and human rights, we identify four new rights that may become of great relevance in the coming decades: the right to cognitive liberty, the right to mental privacy, the right to mental integrity, and the right to psychological continuity.
references:
- Marcello Ienca and Roberto Andorno. Towards new human rights in the age of neuroscience and neurotechnology. Life Sciences, Society and Policy201713:5 DOI: 10.1186/s40504-017-0050-1 (open access)
- Improving Palliative Care with Deep LearningAbstract— Improving the quality of end-of-life care for hospitalized patients is a priority for healthcare organizations. Studies have shown that physicians tend to over-estimate prognoses, which in combination with treatment inertia results in a mismatch between patients wishes and actual care at the end of life . We describe a method to address this problem using Deep Learning and Electronic Health Record (EHR) data, which is currently being piloted, with Institutional Review Board approval, at an academic medical center. The EHR data of admitted patients are automatically evaluated by an algorithm, which brings patients who are likely to benefit from palliative care services to the attention of the Palliative Care team. The algorithm is a Deep Neural Network trained on the EHR data from previous years, to predict all-cause 3-12 month mortality of patients as a proxy for patients that could benefit from palliative care. Our predictions enable the Palliative Care team to take a proactive approach in reaching out to such patients, rather than relying on referrals from treating physicians, or conduct time consuming chart reviews of all patients. We also present a novel interpretation technique which we use to provide explanations of the model’s predictions.I. INTRODUCTION Studies have shown that approximately 80% of Americans would like to spend their final days at home if possible, but only 20% do [1]. In fact, up to 60% of deaths happen in an acute care hospital, with patients receiving aggressive care in their final days. Access to palliative care services in the United States has been on the rise over the past decade. In 2008, 53% of all hospitals with fifty or more beds reported having palliative care teams, rising to 67% in 2015 [2]. However, despite increasing access, data from the National Palliative Care Registry estimates that less than half of the 7-8% of all hospital admissions that need palliative care actually receive it [3]. Though a significant reason for this gap comes from the palliative care workforce shortage [4], and incentives for health systems to employ them, technology can still play a crucial role by efficiently identifying patients who may benefit most from palliative care, but might otherwise be overlooked under current care models. We focus on two aspects of this problem. First, physicians may not refer patients likely to benefit from palliative care for multiple reasons such as overoptimism, time pressures, or treatment inertia [5]. This may lead to patients failing to have their wishes carried out at end of life [6] and overuse of aggressive care. Second, a shortage of palliative care professionals makes proactive identification of candidate patients via manual chart review an expensive and time-consuming process. The criteria for deciding which patients benefit from palliative care can be hard to state explicitly. Our approach uses deep learning to screen patients admitted to the hospital to identify those who are most likely to have palliative care needs. The algorithm addresses a proxy problem - to predict the mortality of a given patient within the next 12 months - and use that prediction for making recommendations for palliative care referral. This frees the palliative care team from manual chart review of every admission and helps counter the potential biases of treating physicians by providing an objective recommendation based on the patient’s EHR. Currently existing tools to identify such patients have limitations, and they are discussed in the next section…. : https://arxiv.org/pdf/1711.06402.pdf
Bioinspired
bio-voltage memristors
Abstract
Memristive devices are
promising candidates to emulate biological computing. However, the typical
switching voltages (0.2-2 V) in previously described devices are much higher
than the amplitude in biological counterparts. Here we demonstrate a type of diffusive
memristor, fabricated from the protein nanowires harvested from the
bacterium Geobacter sulfurreducens, that functions at the
biological voltages of 40-100 mV. Memristive function at biological voltages is
possible because the protein nanowires catalyze metallization. Artificial
neurons built from these memristors not only function at biological action
potentials (e.g., 100 mV, 1 ms) but also exhibit temporal integration close to
that in biological neurons. The potential of using the memristor to directly
process biosensing signals is also demonstrated….: https://www.nature.com/articles/s41467-020-15759-y
AI 2041: Ten
Visions for Our Future
In a
groundbreaking blend of science and imagination, the former president of Google
China and a leading writer of speculative fiction join forces to answer an
urgent question: How will artificial intelligence change our world over the
next twenty years?
AI will be the defining issue of the twenty-first century, but many people know
little about it apart from visions of dystopian robots or flying cars. Though
the term has been around for half a century, it is only now, Kai-Fu Lee argues,
that AI is poised to upend our society, just as the arrival of technologies
like electricity and smart phones did before it. In the past five years, AI has
shown it can learn games like chess in mere hours--and beat humans every time.
AI has surpassed humans in speech and object recognition, even outperforming
radiologists in diagnosing lung cancer. AI is at a tipping point. What comes
next?
Within two decades, aspects of daily life may be unrecognizable. Humankind
needs to wake up to AI, both its pathways and perils. In this provocative work
that juxtaposes speculative storytelling and science, Lee, one of the world's
leading AI experts, has teamed up with celebrated novelist Chen Qiufan to
reveal how AI will trickle down into every aspect of our world by 2041. In ten
gripping short stories that crisscross the globe, coupled with incisive
analysis, Lee and Chen explore AI's challenges and its potential:
- Ubiquitous AI that knows you better than you know yourself
- Genetic fortune-telling that predicts risk of disease or even IQ
- AI sensors that creates a fully contactless society in a future
pandemic
- Immersive personalized entertainment to challenge our notion of
celebrity
- Quantum computing and other leaps that both eliminate and
increase risk
By gazing toward a not-so-distant horizon, AI 2041 offers
powerful insights and compelling storytelling for everyone interested in our
collective future.
https://www.goodreads.com/book/show/56377201-ai-2041
AI 2041: Ten
Visions for Our Future by Kai-Fu
Lee
This inspired
collaboration between a pioneering technologist and a visionary writer of
science fiction offers bold and urgent insights.
People want AI for its
brain power, not its
people skills
Published
on May 4, 2017
Partner at PwC Analytics - Innovation Lead, Artificial
Intelligence Expert
By Anand Rao
Artificial intelligence (“AI”) is fast becoming the
next great democratizer for services. In the medical field, 56% of consumers
surveyed see its potential to lower cost and break down barriers in
providing medical access to lower income adults. And the beginnings of that technology can
already be seen: an AI system has successfully identified autism in babies with 81% accuracy, while a
Stanford-led experiment used AI to identify skin cancer with 91% accuracy.
But as much as these technologies develop and become
more successful in application, the majority of consumers still want a human
touch accompany cutting-edge tech. While consumers trust AI to make vital
decisions on the back end in terms of data processing and analysis, they still
prefer a human to deliver information to them or to help explain a result.
According to our latest Consumer Intelligence Series survey, 77% would prefer to visit a doctor
in person than to take an at-home assessment with a robotic smart kit, and only
22% think it’s likely that people will turn entirely to an AI assistant versus
a human as a doctor.
The same sentiment is echoed for office environments.
AI is great for the processing it brings to the table, but not for making final
decisions. Executives see AI as a liberator when it comes to repetitive tasks
in their day-to-day life. Our survey reveals that paperwork, scheduling and
timesheets seemed like appropriate, tedious tasks to be delegated to a machine.
However, executives were less confident about AI’s ability to handle HR-related
tasks which require a human touch. Sixty-nine percent of executives believe AI
would be just as fair, or even more fair, as a human manager when making
promotion decisions. But in practice, 86% would want to speak with a human
after a review decision made by AI.
Ultimately, people are optimistic that AI will save
time and provide more services to even more people. But when it comes to tasks
that require emotional intelligence, they aren’t ready to hand over the reins
just yet. The emotional skills that humans innately have as, well, humans, are
their strength. Where AI is logical and methodical, humans can be inventive and
empathetic. The role of humans will be to push AI further: to create more
useful programs, to develop more unique ways for machines to think, and to
tackle new problems. All of which will lead to a smarter – both intellectually
and emotionally – world for all.
Man and machine together is better than either one on
their own.
MITCSAIL | Operating Robots with Virtual Reality
Teleoperating robots with virtual reality:
getting inside a robot’s head
October 6, 2017
Researchers at MIT’s Computer
Science and Artificial Intelligence Laboratory (CSAIL) have developed a virtual-reality (VR)
system that lets you teleoperate a robot using an Oculus Rift or HTC Vive VR
headset.
CSAIL’s “Homunculus Model”
system (the classic notion of a small human sitting inside the brain and
controlling the actions of the body) embeds you in a VR control room with
multiple sensor displays, making it feel like you’re inside the robot’s head.
By using gestures, you can control the robot’s matching movements to perform
various tasks.
The system can be connected
either via a wired local network or via a wireless network connection over the
Internet. (The team demonstrated that the system could pilot a robot from
hundreds of miles away, testing it on a hotel’s wireless network in Washington,
DC to control Baxter at MIT.)
According to CSAIL
postdoctoral associate Jeffrey Lipton, lead author on an open-access arXiv paper
about the system (presented this week at the IEEE/RSJ International Conference
on Intelligent Robots and Systems (IROS) in Vancouver), “By teleoperating
robots from home, blue-collar workers would be able to telecommute and benefit
from the IT revolution just as white-collars workers do now.”
Jobs for video-gamers too
The researchers imagine that
such a system could even help employ jobless video-gamers by “game-ifying”
manufacturing positions. (Users with gaming experience had the most ease with
the system, the researchers found in tests.)
To make these movements
possible, the human’s space is mapped into the virtual space, and the virtual
space is then mapped into the robot space to provide a sense of co-location.
The team demonstrated the
Homunculus Model system using the Baxter humanoid robot from Rethink Robotics,
but the approach could work on other robot platforms, the researchers said.
In tests involving pick and
place, assembly, and manufacturing tasks (such as “pick an item and stack it
for assembly”) comparing the Homunculus Model system with existing
state-of-the-art automated remote-control, CSAIL’s Homunculus Model system
had a 100% success rate compared with a 66% success rate for state-of-the-art
automated systems. The CSAIL system was also better at grasping objects 95
percent of the time and 57 percent faster at doing tasks.*
“This contribution
represents a major milestone in the effort to connect the user with the robot’s
space in an intuitive, natural, and effective manner.” says Oussama Khatib, a
computer science professor at Stanford University who was not involved in the
paper.
The team plans to eventually
focus on making the system more scalable, with many users and different types
of robots that are compatible with current automation technologies.
* The Homunculus
Model system solves a delay problem with existing systems, which use a GPU or
CPU, introducing delay. 3D reconstruction from the stereo HD cameras is instead
done by the human’s visual cortex, so the user constantly receives visual
feedback from the virtual world with minimal latency (delay). This also avoids
user fatigue and nausea caused by motion sickness (known as simulator sickness)
generated by “unexpected incongruities, such as delays or relative motions,
between proprioception and vision [that] can lead to the nausea,” the
researchers explain in the paper.
MITCSAIL | Operating Robots with Virtual Reality
Abstract of Baxter’s Homunculus:
Virtual Reality Spaces for Teleoperation in Manufacturing
Expensive specialized
systems have hampered development of telerobotic systems for manufacturing
systems. In this paper we demonstrate a telerobotic system which can reduce the
cost of such system by leveraging commercial virtual reality(VR) technology and
integrating it with existing robotics control software. The system runs on a
commercial gaming engine using off the shelf VR hardware. This system can be
deployed on multiple network architectures from a wired local network to a
wireless network connection over the Internet. The system is based on the
homunculus model of mind wherein we embed the user in a virtual reality control
room. The control room allows for multiple sensor display, dynamic mapping
between the user and robot, does not require the production of duals for the
robot, or its environment. The control room is mapped to a space inside the
robot to provide a sense of co-location within the robot. We compared our
system with state of the art automation algorithms for assembly tasks, showing
a 100% success rate for our system compared with a 66% success rate for
automated systems. We demonstrate that our system can be used for pick and place,
assembly, and manufacturing tasks.
references:
Can
machine-learning improve cardiovascular risk prediction using routine clinical data?
·
Stephen F. Weng ,
·
Jenna Reps ,
·
Joe Kai ,
·
Jonathan M. Garibaldi ,
·
Nadeem Qureshi
- Published: April 4, 2017
- https://doi.org/10.1371/journal.pone.0174944
Current approaches to predict cardiovascular risk fail
to identify many people who would benefit from preventive treatment, while
others receive unnecessary intervention. Machine-learning offers opportunity to
improve accuracy by exploiting complex interactions between risk factors. We
assessed whether machine-learning can improve cardiovascular risk prediction.
Prospective cohort study using routine clinical data
of 378,256 patients from UK family practices, free from cardiovascular disease
at outset. Four machine-learning algorithms (random forest, logistic
regression, gradient boosting machines, neural networks) were compared to an
established algorithm (American College of Cardiology guidelines) to predict
first cardiovascular event over 10-years. Predictive accuracy was assessed by
area under the ‘receiver operating curve’ (AUC); and sensitivity, specificity,
positive predictive value (PPV), negative predictive value (NPV) to predict
7.5% cardiovascular risk (threshold for initiating statins).
24,970 incident cardiovascular events (6.6%) occurred.
Compared to the established risk prediction algorithm (AUC 0.728, 95% CI
0.723–0.735), machine-learning algorithms improved prediction: random forest
+1.7% (AUC 0.745, 95% CI 0.739–0.750), logistic regression +3.2% (AUC 0.760,
95% CI 0.755–0.766), gradient boosting +3.3% (AUC 0.761, 95% CI 0.755–0.766),
neural networks +3.6% (AUC 0.764, 95% CI 0.759–0.769). The highest achieving
(neural networks) algorithm predicted 4,998/7,404 cases (sensitivity 67.5%, PPV
18.4%) and 53,458/75,585 non-cases (specificity 70.7%, NPV 95.7%), correctly
predicting 355 (+7.6%) more patients who developed cardiovascular disease
compared to the established algorithm.
Machine-learning significantly improves accuracy of cardiovascular risk
prediction, increasing the number of patients identified who could benefit from
preventive treatment, while avoiding unnecessary treatment of others.
Figures
Citation: Weng SF, Reps
J, Kai J, Garibaldi JM, Qureshi N (2017) Can machine-learning improve
cardiovascular risk prediction using routine clinical data? PLoS ONE 12(4):
e0174944. https://doi.org/10.1371/journal.pone.0174944
Editor: Bin Liu, Harbin Institute of Technology Shenzhen Graduate School, CHINA
Received: December 14, 2016; Accepted: March 18, 2017; Published: April
4, 2017
Copyright: © 2017 Weng et al. This is an open access article distributed under the
terms of the Creative Commons Attribution License, which
permits unrestricted use, distribution, and reproduction in any medium,
provided the original author and source are credited.
Data Availability: This dataset contains patient level health records with intellectual
property rights held by The Crown copyright, which is subject to UK information
governance laws. The authors will make their data available upon specific
requests subject to the requestor obtaining ethical and research approvals from
the Clinical Practice Research Datalink Independent Scientific Advisory
Committee (https://www.cprd.com/intro.asp)
at the UK Medicines and Health Products Regulatory Agency.
Funding: This paper presents independent research funded by the National Institute
for Health Research School for Primary Care Research (NIHR SPCR): personal
training fellowship award for SW from 2015-2018. URL: https://www.spcr.nihr.ac.uk/trainees.
The views expressed are those of the authors and not necessarily those of the
NIHR, the NHS, or the Department of Health.
Competing interests: The authors have declared that no competing interests exist.
Globally,
cardiovascular disease (CVD) is the leading cause of morbidity and ortality. In
2012, there were 17.5 million deaths from CVD with 7.4 million deaths due to
coronary heart disease (CHD) and 6.7 million deaths due to stroke [1]. Established approaches to CVD risk
assessment, such as that recommended by the American Heart Association/American
College of Cardiology (ACC/AHA), predict future risk of CVD based on
well-established risk factors such as hypertension, cholesterol, age, smoking,
and diabetes. These risk factors have recognised aetiological associations with
CVD and feature within most CVD risk prediction tools (e.g. ACC/AHA [2], QRISK2 [3], Framingham [4], Reynolds [5]. There remain a large number of individuals
at risk of CVD who fail to be identified by these tools, while some individuals
not at risk are given preventive treatment unnecessarily. For instance,
approximately half of myocardial infractions (MIs) and strokes will occur in
people who are not predicted to be at risk of cardiovascular disease [6].
All standard
CVD risk assessment models make an implicit assumption that each risk factor is
related in a linear fashion to CVD outcomes [7]. Such models may thus oversimplify complex
relationships which include large numbers of risk factors with non-linear
interactions. Approaches that better incorporate multiple risk factors, and
determine more nuanced relationships between risk factors and outcomes need to
be explored.
Machine-learning
(ML) offers an alternative approach to standard prediction modelling that may
address current limitations. It has potential to transform medicine by better
exploiting ‘big data’ for algorithm development [7]. ML developed from the study of pattern
recognition and computational learning (so-called ‘artificial intelligence’).
This relies on a computer to learn all complex and non-linear interactions
between variables by minimising the error between predicted and observed
outcomes [8]. In addition to potentially improving
prediction, ML may identify latent variables, which are unlikely to be observed
but might be inferred from other variables [9].
To date, there
has been no large-scale investigation applying machine-learning for prognostic
assessment in the general population, using routine clinical data. The aim of
this study was to evaluate whether machine-learning can improve accuracy of
cardiovascular risk prediction within a large general primary care population.
We also sought to determine which class of machine-learning algorithm has
highest predictive accuracy.
The cohort of patients was derived from the Clinical Practice Research
Datalink (CPRD), anonymized electronic medical records from nearly 700 UK
family practices documenting demographic details, history of medical conditions,
prescription drugs, acute medical outcomes, referrals to specialists,
admissions to hospitals, and biological results. The database is representative
of the UK general population and linked to hospital (secondary care) records [10]. Ethical and research approvals were
granted by the Independent Scientific Advisory Committee (ISAC) at CPRD (number
14_205).
The cohort of patients were registered with a family practice between the
ages of 30 to 84 years at baseline, who had complete data for the eight core
baseline variables (gender, age, smoking status, systolic blood pressure, blood
pressure treatment, total cholesterol, HDL cholesterol, and diabetes) used in
the established ACC/AHA 10-year risk prediction model [2]. The baseline date was set as the 1st of
January 2005, thus allowing all patients within the cohort to be followed-up
for 10 years. The end of the study period was specified as the 1st of
January 2015, the latest date for which CPRD had provided an updated dataset.
Individuals with a previous history of CVD, lipid disorders which are
inherited, prescribed lipid lowering drugs, or outside the specified age range
prior to or on the baseline date were excluded from the analysis.
The eight core risk variables (above) were used to derive a baseline risk
prediction model using the published equations in the 2013 ACC/AHA guidelines
for assessment of CVD risk [2]. To compare the machine-learning
algorithms, an additional 22 variables with potential to be associated with CVD
were included in the analysis. These variables were selected based on their
inclusion in published CVD risk algorithms [2–5], within literature on other potential CVD
risk factors [11–21], and further reviewed by practising
clinicians (NQ, JK).
In nine of the additional continuous variables, there were some levels of
missing data. Median imputation, a common approach to dealing with missing
values in machine-learning algorithms [22] was used. It was also hypothesized that
missing values in certain clinical variables (e.g. BMI and laboratory results)
may indicate a perception of reduced relevance in certain patients, given the
under recording of normal BMI values in primary care medical records [23]. Dummy variables were created to indicate
whether these continuous variable values were missing. For demographic
categorical variables, Townsend deprivation index (28) and ethnicity, missing
values were given a separate category of ‘unknown’ in the analyses. In total,
there were 30 variables (excluding dummy variables for missing values) analysed
in the machine-learning models prior to baseline
The primary outcome was the first recorded diagnosis of a fatal or
non-fatal cardiovascular event documented in the patient’s primary or secondary
care computerised record. In primary care, CVD is labelled and electronically
recorded by UK National Health Service (NHS) Read codes. Further, confirmation
of outcomes in secondary care (Hospital Episodes Statistics) utilised ICD-10
codes, specifically I20 to I25 for coronary (ischaemic) heart conditions and
I60 to I69 for cerebrovascular conditions.
To compare machine-learning risk algorithms, the study population was split
in the data set into a ‘training’ cohort in which the CVD risk algorithms were
derived and a ‘validation’ cohort in which the algorithms were applied and
tested. The ‘training’ cohort was derived from random sampling of 75% of the
extracted CPRD cohort, and the ‘validation’ cohort comprised the remaining 25%.
Four commonly used classes of machine-learning algorithms were utilised:
logistic regression [25], random forest [26], gradient boosting machines [27], and neural networks [28]. These algorithms were selected based on
the ease of implementation into current UK primary care electronic health
records. Development of the risk algorithms in the training cohort and
application of the risk algorithms to the validation cohort was completed using
RStudio with library packages caret (http://CRAN.R-project.org/package=caret)
for neural networks and h2o (http://www.h2o.ai) for the remaining
algorithms. Each model’s hyper parameters were determined by using a grid
search and two fold cross-validation on the training cohort to determine the
values which led to the best performance. Further details on machine-learning
models are described in the S1 Text.
Descriptive characteristic of the study population were provided, including
number (%) and mean (SD) for categorical and continuous variables,
respectively. The performance of the machine-learning prediction algorithms,
developed from the training cohort, was assessed using the validation cohort by
calculating Harrell’s c-statistic [29], a measure of the total area under the
receiver operating characteristic curve (AUC). Standard errors and 95%
confidence intervals were estimated for the c-statistic using a jack-knife
procedure [30]. Additionally, using thresholds
corresponding to the 10-year CVD risk of > 7.5% as recommended by the
ACC/AHA guidelines [2] for initiating lipid lowering therapy,
binary classification analysis was used to compare observed and expected
prediction of cases and non-cases in the validation cohort. This process
provided sensitivity, specificity, positive predictive value (PPV), and
negative predictive value (NPV). The statistical analyses assessing algorithm
performance were performed using STATA 13 MP4.
There were a total of 383,592 patients from 12 million patients in the CPRD
database at baseline (1 Jan 2005) who met eligibility criteria. After excluding
5,336 patients with coding errors (i.e. non-numerical entries for blood
pressure/cholesterol) and extreme outlying observations (> 5 SDs from the
mean), the analysis cohort consisted of 378,256 patients. This cohort was then
randomly split into a 75% sample of 295,267 patients to train the
machine-learning algorithms and the remaining sample of 82,989 patients for
validation
From a total cohort of 378,256 patients who were free from CVD at baseline,
there were 24,970 incident cases (6.6%) of CVD during the 10-year follow-up
period. There were significantly fewer women than men (42% F, 52% M) in CVD
cases while there was only slightly more women than men in non-CVD cases (52%
F, 48% M). The mean baseline age of CVD patients was 65.3 years compared to
57.3 years in non-CVD patients (p < 0.001).
All variables listed in Table 2 were inputs for the
machine-learning models and trained using a cohort of 295,267 patients with
19,487 incident CVD cases (6.6%) of developing over the 10-year follow-up
period. Variable importance was determined by the coefficient effect size for
the ACC/AHA baseline model and machine-learning logistic regression. Random
forest and gradient boosting machine models, based on decision-trees, rank
variable importance on the selection frequency of the variable as a decision
node while neural networks use overall weighting of the variable within the
model.
The standard risk factors in the ACC/AHA algorithm stratified by gender
were age, total cholesterol, HDL cholesterol, smoking, blood pressure, and
diabetes. Several of these risk factors in the ACC/AHA model (age, gender,
smoking) were present as top ranked risk factors for all four machine-learning
algorithms. However, diabetes, which is prominent in many CVD algorithms, was
not present in the top ranked risk factors for any of the machine-learning
models (though HbA1c was included as a proxy in random forest models). Other
new risk factors not found in any previous risk prediction tools but determined
by machine-learning included medical conditions such as COPD and severe mental
illness, prescribing of oral corticosteroids, as well as biomarkers such as
triglyceride levels. Random forest and gradient boosting machines were most
similar in risk factor selection and rankings, with some discrepancies in
ranking order and substitution of BMI for systolic blood pressure. Logistic
regression and neural networks prioritised medical conditions such as atrial
fibrillation, chronic kidney disease, and rheumatoid arthritis over biometric
risk factors. Neural networks also put less weighting on age as a risk factor,
and included ‘BMI missing’ as a protective risk factor of CVD.
Higher
c-statistics results in better algorithm discrimination. The baseline (BL)
ACC/AHA 10-year risk prediction algorithm is provided for comparative purposes.
The ACC/AHA risk model served as a baseline for comparison (AUC 0.728, 95%
CI 0.723–0.735). All machine-learning algorithms tested achieved statistically
significant improvements in discrimination compared to the baseline models
(from 1.7% for random forest algorithms to 3.6% for neural networks)
The ACC/AHA baseline model predicted 4,643 cases correctly from 7,404 total
cases, resulting in a sensitivity of 62.7% and PPV of 17.1%. The random forest
algorithm resulted in a net increase of 191 CVD cases from the baseline model,
increasing the sensitivity to 65.3% and PPV to 17.8% while logistic regression
resulted in a net increase of 324 CVD cases (sensitivity 67.1%; PPV 18.3%).
Gradient boosting machines and neural networks performed best, resulting in a
net increase of 354 (sensitivity 67.5%; PPV 18.4%) and 355 CVD (sensitivity
67.5%; PPV 18.4%) cases correctly predicted, respectively.
The ACC/AHA baseline model correctly predicted 53,106 non-cases from 75,585
total non-cases, resulting in a specificity of 70.3% and NPV of 95.1%. The net
increase in non-cases correctly predicted compared to the baseline ACC/AHA
model ranged from 191 non-cases for the random forest algorithm to 355
non-cases for the neural networks. Full details on classification analysis can
be found in S2 Table.
Compared to an
established AHA/ACC risk prediction algorithm, we found all machine-learning
algorithms tested were better at identifying individuals who will develop CVD
and those that will not. Unlike established approaches to risk prediction, the
machine-learning methods used were not limited to a small set of risk factors,
and incorporated more pre-existing medical conditions. Neural networks
performed the best, with predictive accuracy improving by 3.6%. This is an
encouraging step forward. For example, the addition of emerging biochemical
risk factors, such as high sensitivity C-reactive protein, has recently
achieved less than 1% improvement in CVD risk prediction [31].
To our knowledge, this is the first investigation applying machine-learning
to routine data in patients’ electronic records, demonstrating improved
prediction of CVD risk in a large general population. The study also
illustrates use of a range of machine learning methods, as well as evaluation
techniques, that are lacking in existing applications of machine-learning to
clinical data [32]. Our results are consistent with much
smaller studies [33,34] in more selected populations. For example,
a cohort study of 5,159 men in Northern Germany [34] found a similar 3.2% improvement in
accuracy of prediction of coronary risk using a probabilistic neural network
model.
The current study’s use of an array of machine-learning algorithms has
suggested intriguing variations in the importance of different risk factors
depending on the modelling technique. Models based on decision trees resembled
closely to each other, with gradient boosting machines out-performing random
forests. Neural networks and logistic regression placed far more importance on
categorical variables and CVD-associated medical conditions, clustering
patients with similar characteristics in each groups. This may help inform
further exploration of diverse predictive risk factors, and future development
of new risk prediction approaches and algorithms.
Finally, the importance of missing values or non-response are not often
assessed in development of conventional CVD risk prediction tools [2–5]. This study suggests that missing values,
in particular, for routine biometric variables such as BMI, are independent
predictors of CVD. This is consistent with subjective assessment by clinicians
who may not record normal BMI values if patients appear at lower CVD risk [23].
It is acknowledged that the “black-box” nature of machine-learning
algorithms, in particular neural networks, can be difficult to interpret. This
refers to the inherent complexity in how the risk factor variables are
interacting and their independent effects on the outcome. However, improvements
in data visualization methods have improved understanding of these models,
illustrating the importance of network connections between risk factors [35]
It is also recognised that as the number of potential risk factors
increases, the complexity of the models can cause over-fitting, yielding
implausible results. We addressed this by active and appropriate choice of
pre-training, hyper-parameter selection, and regularisation [36].
Although we have cross-validated the performance of the machine-learning
algorithms using an independent dataset, an approach commonly used for the
development of established cardiovascular risk algorithms applied to clinical
practice [2–5,24,37], it must be acknowledged that the
jack-knife procedure may yield more accurate results as demonstrated in genomic
or proteomic datasets [38,39]. Moreover, these established risk
prediction algorithms for use in clinical practice have been developed from a
binary classification framework which can often result in an unbalanced
dataset. Ensemble learning have been demonstrated as a solution to construct
balanced datasets to enhance prediction performance [40]. These methods are not yet commonplace for
developing risk prediction models in clinical datasets but their utility should
be explored in future studies.
Finally, we note the study was performed in a large cohort of primary care
patients in the UK. However, its demonstration of machine-learning methods, and
use of routine clinical data available within electronic records in several
countries [41], underline applicability to other
populations and health systems.
CVD risk prediction has become increasingly important in clinical
decision-making since the introduction of the recent ACC/AHA and similar
guidelines internationally [2,42]. Machine-learning approaches offer the
exciting prospect of achieving improved and more individualised CVD risk
assessment. This may assist the drive towards personalised medicine, by better
tailoring risk management to individual patients [43,44].
The improvement in predictive accuracy found in the current study should be
further explored using machine learning with other large clinical datasets, in
other populations, and in predicting other disease outcomes. Future
investigation of the feasibility and acceptability of machine-learning
applications in clinical practice will be needed. As the computational capacity
in health care systems is improving, the opportunities to exploit
machine-learning to enhance prediction of disease risk in clinical practice
will become a realistic option [7]. This might increasingly include predicting
protein structure and function from genetic sequences from patients’ clinical
profiles [7]. This will inevitably require exploration
in future studies on utility and clinical applicability other computationally
demanding machine-learning algorithms such as support vector machines and deep
learning for integration into primary care electronic health records. In
several countries, electronic health records across health care organisations
are held on central servers. This may allow new algorithm development to be
performed off-site using cloud computing software, and then returned to the
clinical setting as applications programme interfaces (APIs) for PCs, mobile
devices and tablets.
Compared to an
established risk prediction approach, this study has shown machine-learning
algorithms are better at predicting the absolute number of cardiovascular
disease cases correctly, whilst successfully excluding non-cases. This has been
demonstrated in a large and heterogeneous primary care patient population using
routinely collected electronic health data.
A
generative vision model that trains with high data efficiency and breaks
text-based CAPTCHAs
Science 26
Oct 2017:
Abstract
Abstract
Learning from few examples and generalizing to
dramatically different situations are capabilities of human visual intelligence
that are yet to be matched by leading machine learning models. By drawing
inspiration from systems neuroscience, we introduce a probabilistic generative
model for vision in which message-passing based inference handles recognition,
segmentation and reasoning in a unified way. The model demonstrates excellent
generalization and occlusion-reasoning capabilities, and outperforms deep
neural networks on a challenging scene text recognition benchmark while being
300-fold more data efficient. In addition, the model fundamentally breaks the
defense of modern text-based CAPTCHAs by generatively segmenting characters
without CAPTCHA-specific heuristics. Our model emphasizes aspects like data
efficiency and compositionality that may be important in the path toward
general artificial intelligence.: http://science.sciencemag.org/content/early/2017/10/26/science.aag2612.full
Machine translation of cortical activity
to text with an encoder–decoder framework
Nature Neuroscience (2020)C
Abstract
A
decade after speech was first decoded from human brain signals, accuracy and
speed remain far below that of natural speech. Here we show how to decode the
electrocorticogram with high accuracy and at natural-speech rates. Taking a cue
from recent advances in machine translation, we train a recurrent neural
network to encode each sentence-length sequence of neural activity into an
abstract representation, and then to decode this representation, word by word,
into an English sentence. For each participant, data consist of several spoken
repeats of a set of 30–50 sentences, along with the contemporaneous signals
from ~250 electrodes distributed over peri-Sylvian cortices. Average word error
rates across a held-out repeat set are as low as 3%. Finally, we show how
decoding with limited data can be improved with transfer learning, by training
certain layers of the network under multiple participants’ data…: https://www.nature.com/articles/s41593-020-0608-8
Researchers watch video images people are
seeing, decoded from their fMRI brain scans in near-real-time
Advanced
deep-learning "mind-reading" system even interprets image meaning and
recreates the video images
October 27, 2017
Purdue Engineering
researchers have developed a system that can show what people are seeing in
real-world videos, decoded from their fMRI brain scans — an advanced new form
of “mind-reading” technology that could lead to new insights in brain
function and to advanced AI systems.
The research builds on
previous pioneering research at UC Berkeley’s Gallant Lab, which
created a computer program in 2011 that translated fMRI brain-wave
patterns into images that loosely mirrored a series of images being viewed.
The new system also decodes
moving images that subjects see in videos and does it in near-real-time. But the
researchers were also able to determine the subjects’ interpretations of the
images they saw — for example, interpreting an image as a person or thing — and
could even reconstruct the original images that the subjects saw.
Deep-learning AI system for watching what
the brain sees
Watching in
near-real-time what the brain sees. Visual information generated by a video (a)
is processed in a cascade from the retina through the thalamus (LGN area) to
several levels of the visual cortex (b), detected from fMRI activity patterns
(c) and recorded. A powerful deep-learning technique (d) then models this
detected cortical visual processing. Called a convolutional neural network
(CNN), this model transforms every video frame into multiple layers of
features, ranging from orientations and colors (the first visual layer) to
high-level object categories (face, bird, etc.) in semantic (meaning) space
(the eighth layer). The trained CNN model can then be used to reverse this
process, reconstructing the original videos — even creating new videos that the
CNN model had never watched. (credit: Haiguang Wen et al./Cerebral Cortex)
The researchers acquired
11.5 hours of fMRI data from each of three women subjects watching 972 video
clips, including clips showing people or animals in action and nature scenes.
To decode the fMRI
images, the research pioneered the use of a deep-learning technique called
a convolutional neural network (CNN). The trained CNN
model was able to accurately decode the fMRI blood-flow data to identify
specific image categories. The researchers could compare (in near-real-time)
these viewed video images side-by-side with the computer’s visual
interpretation of what the person’s brain saw.
The researchers were also
able to figure out how certain locations in the visual cortex were associated
with specific information a person was seeing.
Decoding how the visual cortex works
CNNs have been used to
recognize faces and objects, and to study how the brain processes static images
and other visual stimuli. But the new findings represent the first time CNNs
have been used to see how the brain processes videos of natural scenes. This is
“a step toward decoding the brain while people are trying to make sense of
complex and dynamic visual surroundings,” said doctoral student Haiguang Wen.
Wen was first author of
a paper describing the research, appearing online Oct.
20 in the journal Cerebral Cortex.
“Neuroscience is trying to
map which parts of the brain are responsible for specific functionality,” Wen
explained. “This is a landmark goal of neuroscience. I think what we report in
this paper moves us closer to achieving that goal. Using our technique, you may
visualize the specific information represented by any brain location, and
screen through all the locations in the brain’s visual cortex. By doing that,
you can see how the brain divides a visual scene into pieces, and re-assembles
the pieces into a full understanding of the visual scene.”
The researchers also were
able to use models trained with data from one human subject to predict and
decode the brain activity of a different human subject, a process called
“cross-subject encoding and decoding.” This finding is important because it
demonstrates the potential for broad applications of such models to study brain
function, including people with visual deficits.
The research has
been funded by the National Institute of Mental Health. The work is affiliated
with the Purdue Institute for Integrative Neuroscience. Data
reported in this paper are also publicly available at the Laboratory
of Integrated Brain Imaging website.
Abstract of Neural Encoding and
Decoding with Deep Learning for Dynamic Natural Vision
Convolutional neural network
(CNN) driven by image recognition has been shown to be able to explain cortical
responses to static pictures at ventral-stream areas. Here, we further showed
that such CNN could reliably predict and decode functional magnetic resonance
imaging data from humans watching natural movies, despite its lack of any
mechanism to account for temporal dynamics or feedback processing. Using
separate data, encoding and decoding models were developed and evaluated for
describing the bi-directional relationships between the CNN and the brain.
Through the encoding models, the CNN-predicted areas covered not only the
ventral stream, but also the dorsal stream, albeit to a lesser degree;
single-voxel response was visualized as the specific pixel pattern that drove
the response, revealing the distinct representation of individual cortical
location; cortical activation was synthesized from natural images with
high-throughput to map category representation, contrast, and selectivity.
Through the decoding models, fMRI signals were directly decoded to estimate the
feature representations in both visual and semantic spaces, for direct visual
reconstruction and semantic categorization, respectively. These results
corroborate, generalize, and extend previous findings, and highlight the value
of using deep learning, as an all-in-one model of the visual cortex, to
understand and decode natural vision.
references:
Haiguang
Wen, Junxing Shi, Yizhen Zhang, Kun-Han Lu, Jiayue Cao, Zhongming Liu. Neural
Encoding and Decoding with Deep Learning for Dynamic Natural Vision. Cerebral
Cortex, 2017; 1 DOI: 10.1093/cercor/bhx268
Superconducting ‘synapse’ could enable
powerful future neuromorphic supercomputers
Fires 200 million times faster than human
brain, uses one ten-thousandth as much energy
February 7, 2018
A superconducting “synapse”
that “learns” like a biological system, operating like the human brain, has
been built by researchers at the National Institute of Standards and Technology (NIST).
The NIST switch, described
in an open-access paper
in Science Advances, provides a missing link for neuromorphic
(brain-like) computers, according to the researchers. Such “non-von Neumann architecture” future computers could
significantly speed up analysis and decision-making for applications such as
self-driving cars and cancer diagnosis.
The research is supported by
the Intelligence Advanced Research Projects Activity (IARPA)
Cryogenic Computing Complexity Program, which was launched in 2014 with the goal of paving the way to “a
new generation of superconducting supercomputer development beyond the exascale.”*
NIST’s artificial synapse is
a metallic cylinder 10 micrometers in diameter — about 10 times larger than a
biological synapse. It simulates a real synapse by processing incoming
electrical spikes (pulsed current from a neuron) and customizing spiking output
signals. The more firing between cells (or processors), the stronger the
connection. That process enables both biological and artificial synapses to
maintain old circuits and create new ones.
Dramatically faster,
lower-energy-required, compared to human synapses
But the NIST synapse has two
unique features that the researchers say are superior to human synapses and to
other artificial synapses:
- Operating at 100 GHz, it can fire at a rate
that is much faster than the human brain — 1 billion times per second,
compared to a brain cell’s rate of about 50 times per second.
- It uses only about one ten-thousandth as much
energy as a human synapse. The spiking energy is less than 1 attojoule** —
roughly equivalent to the miniscule chemical energy bonding two atoms in a
molecule — compared to the roughly 10 femtojoules (10,000 attojoules) per
synaptic event in the human brain. Current neuromorphic platforms are
orders of magnitude less efficient than the human brain. “We don’t
know of any other artificial synapse that uses less energy,” NIST
physicist Mike Schneider said.
Superconducting devices
mimicking brain cells and transmission lines have been developed, but until
now, efficient synapses — a crucial piece — have been missing. The new
Josephson junction-based artificial synapse would be used in neuromorphic
computers made of superconducting components (which can transmit electricity
without resistance), so they would be more efficient than designs based on
semiconductors or software. Data would be transmitted, processed, and stored in
units of magnetic flux.
The brain is especially
powerful for tasks like image recognition because it processes data both in
sequence and simultaneously and it stores memories in synapses all over
the system. A conventional computer processes data only in sequence and stores
memory in a separate unit.
The new NIST artificial
synapses combine small size, superfast spiking signals, and low energy
needs, and could be stacked into dense 3D circuits for creating large
systems. They could provide a unique route to a far more complex and
energy-efficient neuromorphic system than has been demonstrated with other
technologies, according to the researchers.
Nature
News does raise some concerns about the research, quoting
neuromorphic-technology experts: “Millions of synapses would be necessary
before a system based on the technology could be used for complex computing; it
remains to be seen whether it will be possible to scale it to this level. … The
synapses can only operate at temperatures close to absolute zero, and need to
be cooled with liquid helium. That this might make the chips impractical for
use in small devices, although a large data centre might be able to maintain
them. … We don’t yet understand enough about the key properties of the
[biological] synapse to know how to use them effectively.”
Inside a superconducting synapse
The NIST synapse is a
customized Josephson
junction***, long used in NIST voltage standards. These junctions are a sandwich of
superconducting materials with an insulator as a filling. When an electrical
current through the junction exceeds a level called the critical current,
voltage spikes are produced.
Each artificial synapse uses
standard niobium electrodes but has a unique filling made of nanoscale clusters
(“nanoclusters”) of manganese in a silicon matrix. The nanoclusters — about
20,000 per square micrometer — act like tiny bar magnets with “spins” that can
be oriented either randomly or in a coordinated manner. The number
of nanoclusters pointing in the same direction can be controlled, which
affects the superconducting properties of the junction.
The synapse rests in a
superconducting state, except when it’s activated by incoming current and
starts producing voltage spikes. Researchers apply current pulses in a magnetic
field to boost the magnetic ordering — that is, the number of nanoclusters pointing
in the same direction.
This magnetic effect
progressively reduces the critical current level, making it easier to create a
normal conductor and produce voltage spikes. The critical current is the lowest
when all the nanoclusters are aligned. The process is also reversible: Pulses
are applied without a magnetic field to reduce the magnetic ordering and raise
the critical current. This design, in which different inputs alter the spin
alignment and resulting output signals, is similar to how the brain operates.
Synapse behavior can also be
tuned by changing how the device is made and its operating temperature. By
making the nanoclusters smaller, researchers can reduce the pulse energy needed
to raise or lower the magnetic order of the device. Raising the operating
temperature slightly from minus 271.15 degrees C (minus 456.07 degrees F) to
minus 269.15 degrees C (minus 452.47 degrees F), for example, results in more
and higher voltage spikes.
* Future
exascale supercomputers would run at 1018 exaflops (“flops”
= floating point operations per second) or more. The current fastest
supercomputer — the Sunway TaihuLight — operates at about 0.1 exaflops;
zettascale computers, the next step beyond exascale, would run 10,000 times
faster than that.
** An attojoule
is 10-18 joule, a unit of energy, and is one-thousandth of a
femtojoule.
***
The Josephson effect is the phenomenon of supercurrent
— i.e., a current that flows indefinitely long without any voltage applied
— across a device known as a Josephson junction, which consists of
two superconductors coupled by a weak link. — Wikipedia
Abstract of Ultralow power
artificial synapses using nanotextured magnetic Josephson junctions
Neuromorphic computing
promises to markedly improve the efficiency of certain computational tasks,
such as perception and decision-making. Although software and specialized
hardware implementations of neural networks have made tremendous
accomplishments, both implementations are still many orders of magnitude less
energy efficient than the human brain. We demonstrate a new form of artificial
synapse based on dynamically reconfigurable superconducting Josephson junctions
with magnetic nanoclusters in the barrier. The spiking energy per pulse varies
with the magnetic configuration, but in our demonstration devices, the spiking
energy is always less than 1 aJ. This compares very favorably with the roughly
10 fJ per synaptic event in the human brain. Each artificial synapse is
composed of a Si barrier containing Mn nanoclusters with superconducting Nb
electrodes. The critical current of each synapse junction, which is analogous
to the synaptic weight, can be tuned using input voltage spikes that change the
spin alignment of Mn nanoclusters. We demonstrate synaptic weight training with
electrical pulses as small as 3 aJ. Further, the Josephson plasma frequencies
of the devices, which determine the dynamical time scales, all exceed 100 GHz.
These new artificial synapses provide a significant step toward a neuromorphic
platform that is faster, more energy-efficient, and thus can attain far greater
complexity than has been demonstrated with other technologies.
references:
M.L.
Schneider, C.A. Donnelly, S.E. Russek, B. Baek, M.R. Pufall, P.F. Hopkins, P.D.
Dresselhaus, S. P. Benz and W.H. Rippard. Ultra-low power artificial synapses
using nano-textured magnetic Josephson junctions. Science Advances, 2018 DOI:
10.1126/sciadv.1701329
New material eliminates need for motors or
actuators in future robots, other devices
Low-cost
material could bear 3000 times its own weight, triggered by light or
electricity
June 29, 2018
A “mini arm”
made up of two hinges of actuating nickel hydroxide-oxyhydroxide material
(left) can lift an object 50 times its own weight when triggered (right) by
light or electricity. (credit: University of Hong Kong)
University of Hong Kong
researchers have invented a radical new lightweight material that could replace
traditional bulky, heavy motors or actuators in robots, medical devices,
prosthetic muscles, exoskeletons, microrobots, and other types of devices.
The new actuating material —
nickel hydroxide-oxyhydroxide — can be instantly triggered and wirelessly
powered by low-intensity visible light or electricity at relatively low
intensity. It can exert a force of up to 3000 times its own weight — producing
stress and speed comparable to mammalian skeletal muscles, according to the
researchers.
The material is also
responsive to heat and humidity changes, which could allow autonomous machines
to harness tiny energy changes in the environment.
The major component is
nickel, so the material cost is low, and fabrication uses a simple
electrodeposition process, allowing for scaling up and manufacture in industry.
Developing actuating
materials was identified as the leading grand challenge in “The
grand challenges of Science Robotics” to “deeply root
robotics research in science while developing novel robotic platforms that will
enable new scientific discoveries.”
Using a light
blocker (top) a mini walking bot (bottom) with the “front leg” bent and
straightened alternatively can walk towards a light source. (credit: University
of Hong Kong)
Ref.: Science
Robotics. Source: University of Hong Kong.
November 6, 1999
November 6, 1999
When Ray Kurzweil discusses human destiny, it’s not
always clear whether he’s talking about technology or theology. It’s technology
that defines his resume. He’s spent 50 years inventing ingenious uses for artificial
intelligence.
But like a priest caught playing in a physics lab, he
keeps coming up with inventions inspired by aesthetics and social conscience.
For instance: when he was still in high school, he wrote a program that
composed music — while his latest software writes poetry. In between he created
machines that read print aloud to the blind, software that draws + paints,
electronic keyboards that produce the sounds of acoustic instruments — plus one
of the most advanced, commercially successful forms of computer speech
recognition.
All of these were products of a restless mind consumed
by the question of what will be. A more abstract product of that vision is his
latest book: The Age of the Spiritual Machine: when
computers exceed human intelligence. Kurzweil looks at the exponential
increase in calculating power since the turn of the century. He concludes: in
50 years, machines will not only be smarter than humans — but also smart enough
to persuade us that they are conscious beings.
That assertion has drawn the wrath of several
prominent philosophers, who question his definitions of intelligence +
consciousness. For example: John Searle PhD — the University of California at
Berkeley professor of philosophy — wrote in the New York Review of
Books: “the fatal flaw in Ray Kurzweil’s argument: it rests on the
assumption the main thing humans do is compute. His proposals aren’t science. I
think he got a little carried away and made philosophical errors.”
Debates about how to define intelligence + consciousness
get the most public attention — but a more compelling idea in the book is his
prediction that our children will eventually become human-machine
hybrids. Based on current trends in computer and biological sciences, he claims
super-power intelligence will come from that hybrid. Merging the human body and
computer circuits will enable humanity to re-design itself.
Ray Kurzweil said: “The primary issue isn’t the mass
of the universe — or the possible existence of anti-gravity or of Einstein’s
so-called cosmological constant. The fate of the universe is a decision yet to
be made, one we’ll intelligently consider when the time is right.”
John Searle PhD said: “Kurzweil doesn’t think he’s
writing a book of science fiction. He’s making serious claims that he thinks
are based on solid scientific results.”
The stuff he’s talking about is no less than a
physical hybrid of human beings and their technology. He says the machines
being created today are the beginning of our meta-morphosis from thinking
mammal to all-knowing hybrid.
Biological evolution has already given way to much
more rapid — and less random — technological evolution, Ray Kurzweil argues. He
predicts: within 30 years direct links will be established between the human
brain and computer circuitry. The implications are mind-boggling. Such links
could mean that the entire contents of a brain could be copied (and preserved)
in an external database. Not only would the human brain be supplemented with
enormous amounts of digital memory, it would also be linked to vast information
resources like the internet — at the speed of thought.
That would produce, through direct brain interface: a
virtual reality indistinguishable from objective reality. Ray Kurzweil cites
medical treatments in which silicon chips have been successfully implanted into
human brains. For example: to alleviate symptoms of Parkinson’s disease, or for
cochlear implants for the deaf. He says these are examples of primitive steps
toward his predictions.
While these sorts of visions might seem far-fetched,
other respected futurists find Kurzweil’s ideas compelling. Marvin Minsky PhD —
a well-known professor of media arts + science at the Massachusetts Institute
of Technology — said Kurzweil is a leading futurist of our time.
Kurzweil’s theories are seriously considered — and
that’s evidence of his credentials. Since his teenage years he has been
harnessing computer power to do pattern recognition tasks. He explains that
pattern recognition is: a part of the computer science filed of artificial intelligence
— where we teach computers to recognize abstract patterns, a capability that
dominates human thinking.”
In year 1965, at age 17, his music composing program
won a Westinghouse science award, a visit to the White House and a spot as a
contestant on the old television game show I’ve Got a Secret. His
secret stumped former Miss America Bess Myerson — but was guessed by the second
panelist, the actor Henry Morgan.
By the time he graduated from the Massachusetts
Institute of Technology in 1970, Ray Kurzweil had already achieved his first
business success. He founded a computer database service that helped high
school students choose the right college. He sold it for $100,000 and went on
to create more businesses built on his inventions. Among his best-known is the
Kurzweil reading machine for the blind. It was a true marvel when it was
introduced. CBS news was so impressed with the device that television news
anchor Walter Cronkite used it to deliver his signature sign-off: “… and that’s
the way it was, on Jan. 13, 1976.”
Along the way, he won several accolades in business
and academics, and received many honorary doctorates. In addition to writing
his next book, he’s developing Fat Kat, an artificial intelligence system that
applies algorithms to securities investment decisions.
In Ray Kurzweil’s vision of the future, the
man-machine hybrid won’t happen through a Frankenstein-like amalgam — but
through an elegant technology: micro-scopic, self-replicating robots called
nano-bots that could travel through the human bloodstream and interact with our
body + brain.
Ray Kurzweil said: “The idea is to direct nano-bots to
travel through every capillary in the brain, where they will pass in very close
proximity to every cortex feature. This could enable the tiny machines to scan
each part of the brain and build up a huge database of its contents. And all
these nano-bots could be communicating with each other, such as on a wireless
network. They could also be on the web.”
The breakthrough in nano-technology came several years
ago with the discovery of the nano-tube, a carbon molecule of enormous
strength. Just about anything can be fabricated from nano-tubes: with many
times the strength of conventional materials, but with a fraction of the
weight.
Also, nano-tubes have more capacity for raw computing
power than commonly used silicon. Maybe this combination of features means that
it could be possible: to build machines the size of a human blood cell, that
are programmable with software — maybe even able to self-replicate of
themselves from carbon atoms.
Ray Kurzweil said: “The size of the technology is
shrinking so rapidly, that in 30 years both the size and cost of this scenario
will be feasible.”
Of course, such technology would inevitably be
accompanied by terrifying dangers. By scanning a brain into a database, a
person’s most private thoughts and memories would be vulnerable to intrusions
by hackers. And wouldn’t the brain also be vulnerable to external control of
information, thought processes and even perceptions of reality?
Ray Kurzweil said: “Those are real concerns.
Organizations like governments, religious, or terrorist groups — or just clever
individuals — could put nano-bots in food or water supplies, trillions of them.
These would then make their way inside people and would monitor their thoughts
and even could control them and place them into virtual environments. But we
won’t be defenseless. We have these concerns today at a primitive level with
Trojan horses that make their way into our computers.”
He says there’s no turning back. Once evolution
produced a technological species — humanity — it put us on a relentless quest
for understanding, and control of our universe.
Ray Kurzweil said: “I’m optimistic, but that’s more of
a personal orientation than something I could scientifically argue. There
definitely are dangers, and we do tend to address them imperfectly, so there’s
some possibility this will fail.”
http://www.kurzweilai.net/the-new-york-times-the-soul-of-the-next-new-machine
2040: How
technological innovation will have changed our lives and society
Published
on December 17, 2018 Jeroen Tas
It’s 2040. I’m 80 years old
and fully able to manage my own health. We finally have the secure information
infrastructure that allows us to collect, analyze and compare our health and
behavioral data ourselves. Since all my health data has been digitized and can
be accessed via the cloud, I can work alongside my care team to proactively
manage my health.
Artificial
Intelligence, one of the most hyped topics of 2018, has had an even bigger
impact than the Internet in the late 90's – it changed everything. Deep
insights into what keeps mehealthy are available to me and my
virtual care providers due to huge advances in AI-enabled diagnostic and
therapeutic tools.
Artificial
Intelligence has helped uncover patterns that were previously untraceable. I
rely on my own "digital twin", a virtual 4D version of myself
that covers my complete medical history: genetic, clinical and behavioral. It
is used to accurately predict and simulate my health outcomes. My twin
intervenes if my health is trending in the wrong direction, automatically
recommending corrective actions to help prevent or treat ailments.
My
digital twin is linked to millions of others, forming a gigantic data ocean,
which has created the possibility of large-scale analysis of health
information. This has led to new insights and clinical breakthroughs that
allowed affordable therapies to be developed for diseases that were difficult
to treat in 2018. We now have access to highly personalized therapies,
including genetic engineering.
My digital twin
recently showed the development of a serious heart problem.
My
digital twin recently showed the development of a serious heart problem, that
if treated early enough, is easily preventable. Together with my doctors, I
decided to have a pre-emptive operation. I didn’t worry too much about the
outcome because data shows that almost all people of my age and my condition
have successfully been through similar treatment. The surgery was performed
quickly and accurately by a surgeon who works in a joint human/robot team. The robots automatically adjust
the therapy to the needs of the patient and are constantly self-improving by
exchanging and analyzing treatment data with other machines around the globe. I
walked out of the lab after a couple of hours, knowing that a serious problem
was averted.
Back to 2018. At Philips I’m working
on innovations that have a positive impact on people’s lives and society. It is
one of the reasons why I enjoy coming to work every day. I firmly believe in
the boundless opportunities of innovation. There is much to look forward
to: cheap and rapid genome analysis will bring new therapies and save lives by
detecting diseases earlier. Artificial intelligence, robots and Blockchain will
be the basis of the next wave of automation, and the reduction of substantial
waste in healthcare systems. Virtual reality and chat bots using artificial
intelligence will impact the way we work together with machines.
It's tough to
make predictions, especially about the future.
As
Yogi Berra observed: it's tough to make predictions, especially about the
future. Just think how different it was in 1998 - that was only 20 years ago.
Technology has evolved faster than we could have ever imagined. Back then, only
3% of the world's population was connected to the internet and few Dutch people
saw the utility of a mobile phone.
Many thought that the postal service and fixed phone lines were just fine and
didn’t see any need to be communicating all day long. In reality, smart phones
and apps have changed our behaviors, politics and culture exponentially.
There
will be no straight path from 2018 to 2040. Many of the innovations that will
be part of our daily lives then have yet to be conceptualized. The rate
and extent of adoption of innovation is unpredictable. Moreover, every
technology has unintended consequences. We have recently seen this with the rapid
growth of fake news on social media and its impact on politics. But one thing
is certain: technology will have an ever-greater impact on our society. It is
up to us to ensure that this impact remains positive. We need to involve a
broad spectrum of society to guide the proper application of the technology.
With the right control mechanisms in place and a human view of technology, life
in 2040 will be healthy and worthwhile.
I
love kite surfing and hope to be able to still do it at 80. Whether that will be
the case cannot be predicted. But, innovation -and my friends and family- will
support me in my quest to stay healthy. That much I can predict.
https://www.linkedin.com/pulse/2040-how-technological-innovation-
How Unilever Uses Artificial Intelligence
To Recruit And Train Thousands Of Employees Each Year
It’s hard to live a day in the developed world without
using a Unilever product. The multinational manufactures and distributes over
400 consumer goods brands covering food and beverages, domestic cleaning
products and personal hygiene.
With so many processes to coordinate and manage,
artificial intelligence is quickly becoming essential for organizations of its
scale. This applies to both research and development as well as the huge
support infrastructure needed for a business with 170,000 employees.
Recently it announced that it had developed machine
learning algorithms capable of sniffing your armpit and telling you whether
you are suffering from body odors. While this may seem like "using a
sledgehammer to crack a walnut", the technology which has been developed
could well go on to be used to monitor food for freshness, helping to solve the
problem of food overproduction and waste endemic
in society.
As well as these smart, public-facing initiatives,
though, artificial
intelligence is being put to use behind the scenes to help
screen and assess the more than one million people per year who apply for jobs
with Unilever. If they make the grade and become one of the thousands who are
offered a job, they have AI-powered tools to help them adjust to their new role
and hit the ground running.
AI-enhanced recruiting
Unilever recruits more than 30,000 people a year and
processes around 1.8 million job applications.
This takes a tremendous amount of time and resources.
As a multinational brand operating in 190 countries, applicants are based all
around the world. Finding the right people is an essential ingredient for
success and Unilever can't afford to overlook talent just because it is buried
at the bottom of a pile of CVs.
To tackle this problem, Unilever partnered with
Pymetrics, a specialist in AI recruitment, to create an online platform which
means candidates can be initially assessed from their own homes, in front of a
computer or mobile phone screen.
First, they are asked to play a selection of games
which test their aptitude, logic, and reasoning, and appetite for risk. Machine
learning algorithms are then used to assess their suitability for whatever role
they have applied for, by matching their profiles against those of previously
successful employees.
The second stage of the process involves submitting a
video interview. Again, the assessor is not a human being but a machine
learning algorithm. The algorithm examines the videos of candidates answering
questions for around 30 minutes, and through a mixture of natural language
processing and body language analysis, determines who is likely to be a good
fit.
Unilever's chief of HR, Leena Nair, told me that
around 70,000 person-hours of interviewing and assessing candidates had been
cut, thanks to the automated screening system.
She said "We look for people with a sense of
purpose – systemic thinking, resilience, business acumen. Based on that
profile, the games and the video interview are all programmed to look for cues
in their behavior that will help us understand who will fit in at
Unilever."
Referring to the video interview analytics for their
future leaders program, she tells me “Every screenshot gives us many data
points about the person, so we work with a number of partners and use a lot of
proprietary technology with those partners, and then we select 3,500 or so
people to go through to our discovery center.” After spending a day with real
leaders and recruiters, Unilever selects about 800 people who will be offered a
job.
The system is also designed to give feedback to all
applicants, even those who aren’t successful.
“What I like about the process is that each and every
person who applies to us gets some feedback,” Nair says.
“Normally when people send an application to a
large company it can go into a ‘black hole’ – thank you very much for your CV,
we’ll get back to you – and you never hear from them again.
“All of our applicants get a couple of pages of
feedback, how they did in the game, how they did in the video interviews, what
characteristics they have that fit, and if they don’t fit, the reason why they
didn’t, and what we think they should do to be successful in a future
application.
“It’s an example of artificial intelligence allowing
us to be more human.”
So, while Unilever isn’t quite ready to hand the
entire recruitment process over to machines just yet, it has shown that it can
assist with the initial “sift” when it comes to preliminary screening of
applicants.
Robots to help you settle into the job
After making the grade, another machine
learning-driven initiative is helping new employees get started in their new
roles – adapting to the day-to-day routines as well as the corporate culture at
the business.
Unabot is a natural language processing (NLP) bot
built on Microsoft’s Bot framework, designed to understand what employees need
to know and fetch information for them when it is asked.
“We joke about the fact we don’t know whether it’s a
man or a woman – it’s Unabot,” Nair tells me.
“Unabot doesn’t only answer HR questions, questions
about anything that affects employees should be answered by Unabot, and it is
now the front face for any employee question – they might ask it about IT
systems, or about their allowances – so we are learning about what matters to
employees in real time.”
Through interacting with employees, Unabot has learned
to answer questions such as where parking is available, the timing of shuttle
buses, and when annual salary reviews are due to take place.
Unlike for example Alexa or consumer-facing, customer
service corporate chatbots, Unabot must also be able to filter and apply
information based on who it is speaking to. It is capable of differentiating
the information it passes on based on both the user's geographical location and
their level of seniority within the company.
Unabot was first rolled out for employees based in the
Philippines and is now operating in 36 countries. It has been selected as the
next AI initiative which will be globally rolled out in all of Unilever’s 190
markets.
“It’s a new way of working,” Nair tells me, “We never
go in and say ‘its perfect so let’s roll it out in all countries,’ we learn
what we can in one country and roll it out in the next one.”
Currently, all of its data comes from internal
sources, such as company guidelines, schedules, policy documents and questions
asked by the employees themselves. In the future, this could be expanded to
include external data such as learning materials.
And although it’s early days, the initial analysis
seems to show that the initiative is popular with staff – with 36% of those in
areas where it is deployed having used it at least once, and around 80% going
on to use it again.
One lesson learned early on was that the importance of
providing a frictionless experience.
“So we’ve learned that you have to make anything that
interacts with employees or consumers effortless,” Nair says.
“People interact in different ways – a policy document
is written in a particular way, its three or four pages of what an employee
shouldn’t do. But an employee tends to ask questions in very simplistic ways –
how does this impact my life, where will I find this, what can I do?”
Machine learning – particularly NLP – can overcome
this due to its ability to detect which questions are repeatedly asked, even if
they are asked in different ways, and present the right information.
About Bernard Marr
Bernard
Marr is an internationally best-selling
author, popular keynote speaker, futurist, and a strategic business &
technology advisor to governments and companies. He helps organisations improve
their business performance, use data more intelligently, and understand the
implications of new technologies such as artificial
intelligence, big
data, blockchains,
and the Internet
of Things.
LinkedIn has ranked Bernard as one of the world’s top
5 business influencers. He is a frequent contributor to the World Economic
Forum and writes a regular column for Forbes. Every day Bernard actively
engages his 1.5 million social media followers and shares content that reaches
millions of readers.
Robo-Apocalypse? Not in Your Lifetime
May 21, 2019 J.
BRADFORD DELONG
Not a week goes
by without some new report, book, or commentary sounding the alarm about
technological unemployment and the "future of work." Yet in
considering the threat posed by automation at most levels of the value chain,
we should remember that robots cannot do what humans cannot tell them to do.
BERKELEY – Will the imminent
“rise of the robots” threaten all future human employment? The most thoughtful
discussion of that question can be found in MIT economist David H. Autor’s
2015 paper,
“Why Are There Still so Many Jobs?”, which considers the problem in the context
of Polanyi’s Paradox. Given that “we can know more than we can tell,” the
twentieth-century philosopher Michael Polanyi observed, we shouldn’t assume
that technology can replicate the function of human knowledge itself. Just
because a computer can know everything there is to know about a car doesn’t
mean it can drive it.
CARL BILDT calls on leaders to finalize a new strategic
agenda, from which all policy and personnel matters must follow.
This distinction between tacit
knowledge and information bears directly on the question of what humans will be
doing to produce economic value in the future. Historically, the tasks that
humans have performed have fallen into ten broad categories. The first, and
most basic, is using one’s body to move physical objects, which is followed by
using one’s eyes and fingers to create discrete material goods. The third
category involves feeding materials into machine-driven production processes –
that is, serving as a human robot – which is followed by actually guiding the
operations of a machine (acting as a human microprocessor).
In the fifth and sixth
categories, one is elevated from microprocessor to software, performing
accounting-and-control tasks or facilitating communication and the exchange of
information. In the seventh category, one actually writes the software,
translating tasks into code (here, one encounters the old joke that every
computer needs an additional “Do” command: “Do What I Mean”). In the eighth
category, one provides a human connection, whereas in the ninth, one acts as
cheerleader, manager, or arbiter for other humans. Finally, in the tenth
category, one thinks critically about complex problems, and then devises novel
inventions or solutions to them.
For the past 6,000 years,
tasks in the first category have gradually been offloaded, first to draft
animals and then to machines. For the past 300 years, tasks in the second
category have also been offloaded to machines. In both cases, jobs in
categories three through six – all of which augmented the increasing power of
the machines – became far more prevalent, and wages grew enormously.
But we have since developed
machines that are better than humans at performing tasks in categories three
and four – where we behave like robots and microprocessors – which is why
manufacturing as a share of total employment in advanced economies has
been declining for
two generations, even as the productivity of manufacturing has increased. This
trend, combined with monetary policymakers’ excessive anti-inflationary zeal,
is a major factor contributing to the recent rise of neofascism in the United
States and other Western countries.
Worse, we have now reached
the point where robots are also better than humans at performing the “software”
tasks in categories five and six, particularly when it comes to managing the
flow of information and, it must be said, misinformation. Nonetheless, over the
next few generations, this process of technological development will work
itself out, leaving humans with just four categories of things to do: thinking
critically, overseeing other humans, providing a human connection, and translating
human whims into a language the machines can understand.
The problem is that very few
of us have the genius to produce genuine economic value with our own
creativity. The wealthy can employ only so many personal assistants. And many
cheerleaders, managers, and dispute-settlers are already unnecessary. That
leaves category eight: as long as livelihoods are tied to remunerative
employment, the prospect of preserving a middle-class society will depend on
enormous demand for human connection.
Here, Polanyi’s Paradox
gives us cause for hope. The task of providing “human connection” is not just
inherently emotional and psychological; it also requires tacit knowledge of
social and cultural circumstances that cannot be codified into concrete, routine commands for computers to
follow. Moreover, each advance in technology creates new domains in which tacit
knowledge matters, even when it comes to interacting with the new technologies themselves.
As Autor observes, though
auto manufacturers “employ industrial robots to install windshields …
aftermarket windshield replacement companies employ technicians, not robots.”
It turns out that “removing a broken windshield, preparing the windshield frame
to accept a replacement, and fitting a replacement into that frame demand more
real-time adaptability than any contemporary robot can cost-effectively
approach.” In other words, automation depends on fully controlled conditions,
and humans will never achieve full control of the entire environment.
Some might counter that
artificial-intelligence applications could develop a capacity to absorb “tacit
knowledge.” Yet even if machine-learning algorithms could communicate back to
us why they have made certain decisions, they will only ever work in restricted
environmental domains. The wide range of specific conditions that they need in
order to function properly renders them brittle and fragile, particularly when
compared to the robust adaptability of human beings.
At any rate, if the “rise of
the robots” represents a threat, it won’t be salient within the next two
generations. For now, we should worry less about technological unemployment,
and more about the role of technology in spreading disinformation. Without a
properly functioning public
sphere, why bother debating economics in the first place?
J. Bradford DeLong is Professor of Economics at the University of
California at Berkeley and a research associate at the National Bureau of
Economic Research. He was Deputy Assistant US Treasury Secretary during the
Clinton Administration, where he was heavily involved in budget and trade
negotiations. His role in designing the bailout of Mexico during the 1994 peso
crisis placed him at the forefront of Latin America’s transformation into a
region of open economies, and cemented his stature as a leading voice in
economic-policy debates.
https://www.project-syndicate.org/commentary/rise-of-robots-social-work-by-j-bradford-delong-2019-05?utm_
Being human in the age of artificial
intelligence
by
Max Tegmark PhD, 2017
This
book revolves around author, physicist, and cosmologist Max Tegmark PhD.
He sets-out to differentiate the myths of AI from reality in an approachable
way. He manages to answer challenging questions about creating a
prosperous world with AI — and how to protect AI from being hacked.
The
book aims to help the layperson understand what will be most affected by AI in
our day-to-day lives. A great primer into the world of AI.
https://s3.amazonaws.com/arena-attachments/1446178/cffa5ebc74cee2b1edf58fa9a5bbcb1c.pdf?1511265314
Artificial
Intelligence Set: What You Need to Know About AI
April 25, 2018
What do you really need to know about the
Artificial Intelligence (AI) revolution? This specially priced 4 item set will
make it easier for you to understand how your company, industry, and career can
be transformed by AI. It is a must-have for managers who need to recognize the
potential impact of AI, how it is driving future growth, and how they can make
the most of it. This collection includes: "Human + Machine: Reimagining
Work in the Age of AI" by Paul Daugherty and H. James Wilson;
which reveals how companies are using the new rules of AI to leap ahead on
innovation and profitability, as well as what you can do to achieve similar
results. Based on the authors' experience and research with 1,500
organizations, this book describes six new types of hybrid human + machine
roles that every company must develop, and it includes a "leader's guide"
with the principals required to become an AI-fueled business. "Prediction
Machines: The Simple Economics of Artificial Intelligence" by Ajay
Agrawal, Joshua Gans, and Avi Goldfarb; the authors lift the curtain on
the AI-is-magic hype and show how basic tools from economics provide clarity
about the AI revolution and a basis for action by CEOs, managers, policy
makers, investors, and entrepreneurs. "Artificial Intelligence for the
Real World" (Article PDF), based on a survey of 250 executives
familiar with their companies' use of cognitive technology and a study of 152
projects show that companies do better by developing an incremental approach to
AI, and by focusing on augmenting rather than replacing human capabilities. And
"Reshaping Business with Artificial Intelligence" (Article
PDF); provides baseline information on the strategies used by companies leading
in AI, the prospects for its growth, and the steps executives need to take to
develop a strategy for their business.
Human + Machine re-Imagining work
in the age of AI
by
Paul Daugherty + H. James Wilson, 2018
Look
around you. Artificial intelligence is no longer just a futuristic notion. It's
here right now--in software that senses what we need, supply chains that
"think" in real time, and robots that respond to changes in their
environment. Twenty-first-century pioneer companies are already using AI to
innovate and grow fast. The bottom line is this: Businesses that
understand how to harness AI can surge ahead. Those that neglect it will fall
behind. Which side are you on?
In Human
+ Machine, Accenture leaders Paul R. Daugherty and H. James (Jim) Wilson
show that the essence of the AI paradigm shift is the transformation of
all business processes within an organization--whether related
to breakthrough innovation, everyday customer service, or personal productivity
habits. As humans and smart machines collaborate ever more closely, work
processes become more fluid and adaptive, enabling companies to change them on
the fly--or to completely reimagine them. AI is changing all the rules of how
companies operate.
Based
on the authors' experience and research with 1,500 organizations, the book
reveals how companies are using the new rules of AI to leap ahead on innovation
and profitability, as well as what you can do to achieve similar results. It
describes six entirely new types of hybrid human + machine roles that every
company must develop, and it includes a "leader’s guide" with the
five crucial principles required to become an AI-fueled business.
Human + Machine provides the
missing and much-needed management playbook for success in our new age of AI.
Our Final Invention: Artificial
Intelligence And the end of the human era.
by James Barrat, 2013
by James Barrat, 2013
In
as little as a decade, artificial intelligence could match, then surpass human
intelligence. Corporations & government agencies around the world are
pouring billions into achieving AI’s Holy Grail—human-level intelligence. Once
AI has attained it, scientists argue, it will have survival drives much like
our own. We may be forced to compete with a rival more cunning, more powerful
& more alien than we can imagine. Thru profiles of tech visionaries,
industry watchdogs & groundbreaking AI systems, James Barrat's Our
Final Invention explores the perils of the heedless pursuit of
advanced AI. Until now, human intelligence has had no rival. Can we coexist
with beings whose intelligence dwarfs our own? Will they allow us to?
SuperIntelligence paths, dangers,
strategies
by Nick Bostrom PhD, 2014
by Nick Bostrom PhD, 2014
The
human brain has some capabilities that the brains of other animals lack. It is
to these distinctive capabilities that our species owes its dominant position.
Other animals have stronger muscles or sharper claws, but we have cleverer
brains.
If machine brains one day come to surpass human brains in general intelligence, then this new superintelligence could become very powerful. As the fate of the gorillas now depends more on us humans than on the gorillas themselves, so the fate of our species then would come to depend on the actions of the machine superintelligence.
But we have one advantage: we get to make the first move. Will it be possible to construct a seed AI or otherwise to engineer initial conditions so as to make an intelligence explosion survivable? How could one achieve a controlled detonation?
To get closer to an answer to this question, we must make our way through a fascinating landscape of topics and considerations. Read the book and learn about oracles, genies, singletons; about boxing methods, tripwires, and mind crime; about humanity's cosmic endowment and differential technological development; indirect normativity, instrumental convergence, whole brain emulation and technology couplings; Malthusian economics and dystopian evolution; artificial intelligence, and biological
cognitive enhancement, and collective intelligence.
This profoundly ambitious and original book picks its way carefully through a vast tract of forbiddingly difficult intellectual terrain. Yet the writing is so lucid that it somehow makes it all seem easy. After an utterly engrossing journey that takes us to the frontiers of thinking about the human condition and the future of intelligent life, we find in Nick Bostrom's work nothing less than a reconceptualization of the essential task of our time.
If machine brains one day come to surpass human brains in general intelligence, then this new superintelligence could become very powerful. As the fate of the gorillas now depends more on us humans than on the gorillas themselves, so the fate of our species then would come to depend on the actions of the machine superintelligence.
But we have one advantage: we get to make the first move. Will it be possible to construct a seed AI or otherwise to engineer initial conditions so as to make an intelligence explosion survivable? How could one achieve a controlled detonation?
To get closer to an answer to this question, we must make our way through a fascinating landscape of topics and considerations. Read the book and learn about oracles, genies, singletons; about boxing methods, tripwires, and mind crime; about humanity's cosmic endowment and differential technological development; indirect normativity, instrumental convergence, whole brain emulation and technology couplings; Malthusian economics and dystopian evolution; artificial intelligence, and biological
cognitive enhancement, and collective intelligence.
This profoundly ambitious and original book picks its way carefully through a vast tract of forbiddingly difficult intellectual terrain. Yet the writing is so lucid that it somehow makes it all seem easy. After an utterly engrossing journey that takes us to the frontiers of thinking about the human condition and the future of intelligent life, we find in Nick Bostrom's work nothing less than a reconceptualization of the essential task of our time.
the Singularity Is Near When humans transcend
biology.
by Ray Kurzweil
by Ray Kurzweil
The
book builds on the ideas introduced in Kurzweil's previous books, The Age of Intelligent Machines (1990)
and The Age of Spiritual Machines (1999).
This time, however, Kurzweil embraces the term the Singularity, which was
popularized by Vernor
Vinge in his 1993 essay "The Coming Technological
Singularity" more than a decade earlier.[1]
Kurzweil
describes his law of accelerating returns which
predicts an exponential increase in technologies like computers, genetics, nanotechnology, robotics and artificial intelligence. Once the
Singularity has been reached, Kurzweil says that machine intelligence will be
infinitely more powerful than all human intelligence combined. Afterwards he
predicts intelligence will radiate outward from the planet until it saturates
the universe. The Singularity is also the point at which machines intelligence
and humans would merge.
Machine Learning Yearning Technical strategy for AI
engineers in the era of deep learning.
by
Andrew Ng PhD, 2019
AI,
machine learning, and deep learning are transforming numerous industries. But
building a machine learning system requires that you make practical decisions:
Should you collect more training data?
Should you use end-to-end deep learning?
How do you deal with your training set not matching your test set?
and many more.
Historically, the only way to learn how to make these "strategy" decisions has been a multi-year apprenticeship in a graduate program or company. This is a book to help you quickly gain this skill, so that you can become better at building AI systems.
Should you collect more training data?
Should you use end-to-end deep learning?
How do you deal with your training set not matching your test set?
and many more.
Historically, the only way to learn how to make these "strategy" decisions has been a multi-year apprenticeship in a graduate program or company. This is a book to help you quickly gain this skill, so that you can become better at building AI systems.
The Master Algorithm How the quest for
the ultimate learning machine will re-make our world.
by
Pedro Domingos, 2015
A
thought-provoking and wide-ranging exploration of machine learning and the race
to build computer intelligences as flexible as our own
In the world's top research labs and universities, the race is on to invent the ultimate learning algorithm: one capable of discovering any knowledge from data, and doing anything we want, before we even ask. In The Master Algorithm, Pedro Domingos lifts the veil to give us a peek inside the learning machines that power Google, Amazon, and your smartphone. He assembles a blueprint for the future universal learner--the Master Algorithm--and discusses what it will mean for business, science, and society. If data-ism is today's philosophy, this book is its bible.
In the world's top research labs and universities, the race is on to invent the ultimate learning algorithm: one capable of discovering any knowledge from data, and doing anything we want, before we even ask. In The Master Algorithm, Pedro Domingos lifts the veil to give us a peek inside the learning machines that power Google, Amazon, and your smartphone. He assembles a blueprint for the future universal learner--the Master Algorithm--and discusses what it will mean for business, science, and society. If data-ism is today's philosophy, this book is its bible.
Rise
of the Robots Technology and the threat of a jobless future.
by
Martin Ford, 2015
What
are the jobs of the future? How many will there be? And who will have them? We
might imagine—and hope—that today's industrial revolution will unfold like the
last: even as some jobs are eliminated, more will be created to deal with the
new innovations of a new era. In Rise of the Robots, Silicon Valley
entrepreneur Martin Ford argues that this is absolutely not the case. As
technology continues to accelerate and machines begin taking care of
themselves, fewer people will be necessary. Artificial intelligence is already
well on its way to making “good jobs” obsolete: many paralegals, journalists,
office workers, and even computer programmers are poised to be replaced by
robots and smart software. As progress continues, blue and white collar jobs
alike will evaporate, squeezing working- and middle-class families ever
further. At the same time, households are under assault from exploding costs,
especially from the two major industries—education and health care—that, so
far, have not been transformed by information technology. The result could well
be massive unemployment and inequality as well as the implosion of the consumer
economy itself.
In Rise of the Robots, Ford details what machine intelligence and robotics can accomplish, and implores employers, scholars, and policy makers alike to face the implications. The past solutions to technological disruption, especially more training and education, aren't going to work, and we must decide, now, whether the future will see broad-based prosperity or catastrophic levels of inequality and economic insecurity. Rise of the Robots is essential reading for anyone who wants to understand what accelerating technology means for their own economic prospects—not to mention those of their children—as well as for society as a whole.
https://www.goodreads.com/book/show/22928874-rise-of-the-robotsIn Rise of the Robots, Ford details what machine intelligence and robotics can accomplish, and implores employers, scholars, and policy makers alike to face the implications. The past solutions to technological disruption, especially more training and education, aren't going to work, and we must decide, now, whether the future will see broad-based prosperity or catastrophic levels of inequality and economic insecurity. Rise of the Robots is essential reading for anyone who wants to understand what accelerating technology means for their own economic prospects—not to mention those of their children—as well as for society as a whole.
Artificial Intelligence. Structures and Strategies for
Complex Problem Solving
by George F Luger
AGI
Ruin: A List of Lethalities
(«Порча сильного ИИ: список смертельных
опасностей»)
Preamble:
(If you're already
familiar with all basics and don't want any preamble, skip ahead to Section
B for technical difficulties of alignment proper.)
I have several times
failed to write up a well-organized list of reasons why AGI will kill
you. People come in with different ideas about why AGI would be
survivable, and want to hear different obviously key points
addressed first. Some fraction of those people are loudly upset with me
if the obviously most important points aren't addressed immediately, and I
address different points first instead.
Having failed to solve
this problem in any good way, I now give up and solve it poorly with a poorly
organized list of individual rants. I'm not particularly happy with this
list; the alternative was publishing nothing, and publishing this seems
marginally more dignified.
Three points about the
general subject matter of discussion here, numbered so as not to conflict with
the list of lethalities:
-3. I'm assuming you are already familiar with
some basics, and already know what 'orthogonality' and 'instrumental
convergence' are and why they're true. People occasionally
claim to me that I need to stop fighting old wars here, because, those people
claim to me, those wars have already been won within the
important-according-to-them parts of the current audience. I suppose it's
at least true that none of the current major EA funders seem to be visibly in denial
about orthogonality or instrumental convergence as such; so, fine. If you
don't know what 'orthogonality' or 'instrumental convergence' are, or don't see
for yourself why they're true, you need a different introduction than this one.
-2. When I say that alignment is lethally
difficult, I am not talking about ideal or perfect goals of 'provable'
alignment, nor total alignment of superintelligences on exact human values, nor
getting AIs to produce satisfactory arguments about moral dilemmas which
sorta-reasonable humans disagree about, nor attaining an absolute certainty of
an AI not killing everyone. When I say that alignment is difficult, I
mean that in practice, using the techniques we actually have, "please
don't disassemble literally everyone with probability roughly 1" is an
overly large ask that we are not on course to get. So far as I'm
concerned, if you can
get a powerful AGI that carries out some pivotal superhuman engineering task,
with a less than fifty percent change of killing more than one billion people,
I'll take it. Even smaller chances of killing even fewer people would be
a nice luxury, but if you can get as incredibly far as "less than roughly
certain to kill everybody", then you can probably get down to under a 5%
chance with only slightly more effort. Practically all of the difficulty
is in getting to "less than certainty of killing literally
everyone". Trolley problems are not an interesting subproblem in all
of this; if there are any survivors, you solved alignment. At this point,
I no longer care how it works, I don't care how you got there, I am
cause-agnostic about whatever methodology you used, all I am looking at is
prospective results, all I want is that we have justifiable cause to believe of
a pivotally useful AGI 'this will not kill literally everyone'. Anybody
telling you I'm asking for stricter 'alignment' than this has failed at reading
comprehension. The big ask from AGI alignment, the basic challenge I am
saying is too difficult, is to obtain by any strategy whatsoever a significant
chance of there being any survivors.
-1. None of this is about anything being
impossible in principle. The metaphor I usually use is that if a textbook
from one hundred years in the future fell into our hands, containing all of the
simple ideas that actually work robustly in practice, we could
probably build an aligned superintelligence in six months. For people
schooled in machine learning, I use as my metaphor the difference between ReLU
activations and sigmoid activations. Sigmoid activations are complicated
and fragile, and do a terrible job of transmitting gradients through many
layers; ReLUs are incredibly simple (for the unfamiliar, the activation function
is literally max(x, 0)) and work much better. Most neural networks for
the first decades of the field used sigmoids; the idea of ReLUs wasn't
discovered, validated, and popularized until decades later. What's lethal
is that we do not have the Textbook From The Future telling us
all the simple solutions that actually in real life just work and are robust;
we're going to be doing everything with metaphorical sigmoids on the first
critical try. No difficulty discussed here about AGI alignment is claimed
by me to be impossible - to merely human science and engineering, let alone in
principle - if we had 100 years to solve it using unlimited retries, the way
that science usually has an unbounded time budget and
unlimited retries. This list of lethalities is about things we
are not on course to solve in practice in time on the first critical try; none
of it is meant to make a much stronger claim about things that are impossible
in principle.
That said:
Here, from my
perspective, are some different true things that could be said, to contradict
various false things that various different people seem to believe, about why
AGI would be survivable on anything remotely remotely resembling the current
pathway, or any other pathway we can easily jump to….:
https://www.lesswrong.com/posts/uMQ3cqWDPHhjtiesc/agi-ruin-a-list-of-lethalities
Pause Giant AI
Experiments
An Open Letter
We
call on all AI labs to immediately pause for at least 6 months the training of
AI systems more powerful than GPT-4.
March 22, 2023
AI systems with
human-competitive intelligence can pose profound risks to society and humanity,
as shown by extensive research[1] and acknowledged by top AI
labs.[2] As stated in the widely-endorsed Asilomar AI
Principles, Advanced AI could represent a profound change in the
history of life on Earth, and should be planned for and managed with
commensurate care and resources. Unfortunately, this level of planning and
management is not happening, even though recent months have seen AI labs locked
in an out-of-control race to develop and deploy ever more powerful digital
minds that no one – not even their creators – can understand, predict, or
reliably control.
Contemporary AI systems
are now becoming human-competitive at general tasks,[3] and we
must ask ourselves: Should we let machines flood our
information channels with propaganda and untruth? Should we
automate away all the jobs, including the fulfilling ones? Should we
develop nonhuman minds that might eventually outnumber, outsmart, obsolete
and replace us? Should we risk loss of control of our
civilization? Such decisions must not be delegated to unelected tech
leaders. Powerful AI systems should be developed only once we are
confident that their effects will be positive and their risks will be
manageable. This confidence must be well justified and increase with
the magnitude of a system's potential effects. OpenAI's recent
statement regarding artificial general intelligence, states that "At
some point, it may be important to get
independent review before starting to train future systems, and for the most
advanced efforts to agree to limit the rate of growth of compute used for
creating new models." We
agree. That point is now.
Therefore, we call
on all AI labs to immediately pause for at least 6 months the training of AI
systems more powerful than GPT-4. This pause should be public and
verifiable, and include all key actors. If such a pause cannot be enacted
quickly, governments should step in and institute a moratorium.
AI labs and independent
experts should use this pause to jointly develop and implement a set of shared
safety protocols for advanced AI design and development that are rigorously
audited and overseen by independent outside experts. These protocols should
ensure that systems adhering to them are safe beyond a reasonable doubt.[4] This
does not mean a pause on AI development in general, merely a stepping
back from the dangerous race to ever-larger unpredictable black-box models with
emergent capabilities.
AI research and
development should be refocused on making today's powerful, state-of-the-art
systems more accurate, safe, interpretable, transparent, robust, aligned,
trustworthy, and loyal.
In parallel, AI
developers must work with policymakers to dramatically accelerate development
of robust AI governance systems. These should at a minimum include: new and
capable regulatory authorities dedicated to AI; oversight and tracking of
highly capable AI systems and large pools of computational capability;
provenance and watermarking systems to help distinguish real from synthetic and
to track model leaks; a robust auditing and certification ecosystem; liability
for AI-caused harm; robust public funding for technical AI safety research; and
well-resourced institutions for coping with the dramatic economic and political
disruptions (especially to democracy) that AI will cause.
Humanity can enjoy a
flourishing future with AI. Having succeeded in creating powerful AI systems,
we can now enjoy an "AI summer" in which we reap the rewards,
engineer these systems for the clear benefit of all, and give society a chance
to adapt. Society has hit pause on other technologies with potentially
catastrophic effects on society.[5] We can do so
here. Let's enjoy a long AI summer, not rush unprepared into a fall.
Signatures
27565
.S. We have prepared some FAQs in response to questions and discussion in the media and elsewhere. You can find them here : https://futureoflife.org/wp-content/uploads/2023/04/FLI_Policymaking_In_The_Pause.pdf
Policymaking in the Pause
What
can policymakers do now to combat risks from advanced AI systems?
12th April 2023
“We don’t know what these
[AI] systems are trained on or how they are being built. All of this happens
behind closed doors at commercial companies. This is worrying.” Catelijne
Muller, President of ALLAI, Member of the EU High Level Expert Group on AI “It
feels like we are moving too quickly. I think it is worth getting a little bit
of experience with how they can be used and misused before racing to build the
next one. This shouldn’t be a race to build the next model and get it out
before others.” Peter Stone, Professor at the University of Texas at Austin,
Chair of the One Hundred Year Study on AI. “Those making these [AI systems]
have themselves said they could be an existential threat to society and even
humanity, with no plan to totally mitigate these risks. It is time to put
commercial priorities to the side and take a pause for the good of everyone to
assess rather than race to an uncertain future” Emad Mostaque, Founder and CEO
of Stability AI “We have a perfect storm of corporate irresponsibility,
widespread adoption, lack of regulation and a huge number of unknowns. [FLI’s
Letter] shows how many people are deeply worried about what is going on. I
think it is a really important moment in the history of AI - and maybe
humanity,” Gary Marcus, Professor Emeritus of Psychology and Neural Science at
New York University, Founder of Geometric Intelligence “The time for saying
that this is just pure research has long since passed. […] It’s in no country’s
interest for any country to develop and release AI systems we cannot control.
Insisting on sensible precautions is not anti-industry. Chernobyl destroyed
lives, but it also decimated the global nuclear industry. I’m an AI researcher.
I do not want my field of research destroyed. Humanity has much to gain from
AI, but also everything to lose.” Stuart Russell, Smith-Zadeh Chair in
Engineering and Professor of Computer Science at the University of California,
Berkeley, Founder of the Center for HumanCompatible Artificial Intelligence
(CHAI). “Let’s slow down. Let’s make sure that we develop better guardrails,
let’s make sure that we discuss these questions internationally just like we’ve
done for nuclear power and nuclear weapons. Let’s make sure we better
understand these very large systems, that we improve on their robustness and
the process by which we can audit them and verify that they are safe for the
public.” Yoshua Bengio, Scientific Director of the Montreal Institute for
Learning Algorithms (MILA), Professor of Computer Science and Operations
Research at the Université de Montréal, 2018 ACM A.M. Turing Award Winner.
FUTURE OF LIFE INSTITUTE 3 CONTENTS 4 Introduction 5 Policy recommendations 6
Mandate robust third-party auditing and certification for specific AI systems 7
Regulate organizations’ access to computational power 8 Establish capable AI
agencies at national level 9 Establish liability for AI-caused harm 10
Introduce measures to prevent and track AI model leaks 10 Expand technical AI
safety research funding 11 Develop standards for identifying and managing
AIgenerated content and recommendations 14 Conclusion FUTURE OF LIFE INSTITUTE
4 Introduction Prominent AI researchers have identified a range of dangers that
may arise from the present and future generations of advanced AI systems if
they are left unchecked. AI systems are already capable of creating
misinformation and authentic-looking fakes that degrade the shared factual
foundations of society and inflame political tensions.1 AI systems already show
a tendency toward amplifying entrenched discrimination and biases, further
marginalizing disadvantaged communities and diverse viewpoints.2 The current,
frantic rate of development will worsen these problems significantly. As these
types of systems become more sophisticated, they could destabilize labor
markets and political institutions, and lead to the concentration of enormous
power in the hands of a small number of unelected corporations. Advanced AI
systems could also threaten national security, e.g., by facilitating the
inexpensive development of chemical, biological, and cyber weapons by non-state
groups. The systems could themselves pursue goals, either human- or
self-assigned, in ways that place negligible value on human rights, human
safety, or, in the most harrowing scenarios, human existence.3 In an eort to
stave o these outcomes, the Future of Life Institute (FLI), joined by over
20,000 leading AI researchers, professors, CEOs, engineers, students, and
others on the frontline of AI progress, called for a pause of at least six
months on the riskiest and most resourceintensive AI experiments – those
experiments seeking to further scale up the size and general capabilities of
the most powerful systems developed to date.4 The proposed pause provides time
to better understand these systems, to reflect on their ethical, social, and
safety implications, and to ensure that AI is developed and used in a
responsible manner. The unchecked competitive dynamics in the AI industry
incentivize aggressive development at the expense of caution5 . In contrast to
the breakneck pace of development, however, the levers of governance are
generally slow and deliberate. A pause on the production of even more powerful
AI systems would thus provide an important opportunity for the instruments of
governance to catch up with the rapid evolution of the field. We have called on
AI labs to institute a development pause until they have protocols in place to
ensure that their systems are safe beyond a reasonable doubt, for individuals,
communities, and society. Regardless of whether the labs will heed our call,
this policy brief provides policymakers with concrete recommendations for how
governments can manage AI risks. The recommendations are by no means
exhaustive: the project of AI governance is perennial 1 See, e.g., Steve
Rathje, Jay J. Van Bavel, & Sander van der Linden, ‘Out-group animosity
drives engagement on social media,’ Proceedings of the National Academy of
Sciences, 118 (26) e2024292118, Jun. 23, 2021, and Tiany Hsu & Stuart A.
Thompson, ‘Disinformation Researchers Raise Alarms About A.I. Chatbots,’ The
New York Times, Feb. 8, 2023 [upd. Feb. 13, 2023] 2 See, e.g., Abid, A.,
Farooqi, M. and Zou, J. (2021a), ‘Large language models associate Muslims with
violence’, Nature Machine Intelligence, Vol. 3, pp. 461–463. 3 In a 2022 survey
of over 700 leading AI experts, nearly half of respondents gave at least a 10%
chance of the long-run eect of advanced AI on humanity being ‘extremely bad,’
at the level of ‘causing human extinction or similarly permanent and severe
disempowerment of the human species.’ 4 Future of Life Institute, ‘Pause Giant
AI Experiments: An Open Letter,’ Mar. 22, 2023. 5 Recent news about AI labs
cutting ethics teams suggests that companies are failing to prioritize the
necessary safeguards. FUTURE OF LIFE INSTITUTE 5 and will extend far beyond any
pause. Nonetheless, implementing these recommendations, which largely reflect a
broader consensus among AI policy experts, will establish a strong governance
foundation for AI. Policy recommendations: 1. Mandate robust third-party
auditing and certification. 2. Regulate access to computational power. 3.
Establish capable AI agencies at the national level. 4. Establish liability for
AI-caused harms. 5. Introduce measures to prevent and track AI model leaks. 6.
Expand technical AI safety research funding. 7. Develop standards for
identifying and managing AI-generated content and recommendations. To coordinate,
collaborate, or inquire regarding the recommendations herein, please contact us
at policy@futureoflife.org. FUTURE OF LIFE INSTITUTE 6 1. Mandate robust
third-party auditing and certification for specific AI systems For some types
of AI systems, the potential to impact the physical, mental, and financial
wellbeing of individuals, communities, and society is readily apparent. For
example, a credit scoring system could discriminate against certain ethnic
groups. For other systems – in particular general-purpose AI systems6 – the
applications and potential risks are often not immediately evident.
General-purpose AI systems trained on massive datasets also have unexpected
(and often unknown) emergent capabilities.7 In Europe, the draft AI Act already
requires that, prior to deployment and upon any substantial modification,
‘high-risk’ AI systems undergo ‘conformity assessments’ in order to certify
compliance with specified harmonized standards or other common specifications.8
In some cases, the Act requires such assessments to be carried out by
independent third-parties to avoid conflicts of interest. In contrast, the
United States has thus far established only a general, voluntary framework for
AI risk assessment.9 The National Institute of Standards and Technology (NIST),
in coordination with various stakeholders, is developing so-called ‘profiles’
that will provide specific risk assessment and mitigation guidance for certain
types of AI systems, but this framework still allows organizations to simply ‘accept’
the risks that they create for society instead of addressing them. In other
words, the United States does not require any third-party risk assessment or
risk mitigation measures before a powerful AI system can be deployed at scale.
To ensure proper vetting of powerful AI systems before deployment, we recommend
a robust independent auditing regime for models that are general-purpose,
trained on large amounts of compute, or intended for use in circumstances
likely to impact the rights or the wellbeing of individuals, communities, or
society. This mandatory third-party auditing and certification scheme could be
derived from the EU’s proposed ‘conformity assessments’ and should be adopted
by jurisdictions worldwide10 . In particular, we recommend third-party auditing
of such systems across a range of benchmarks for the assessment of risks11,
including possible weaponization12 and unethical behaviors13 and mandatory
certification by accredited third-party auditors before these high-risk systems
can be deployed. Certification should only be granted if the developer of the
system can demonstrate that appropriate measures have been taken to mitigate
risk, and that any 6 The Future of Life Institute has previously defined
“general-purpose AI system” to mean ‘an AI system that can accomplish or be
adapted to accomplish a range of distinct tasks, including some for which it
was not intentionally and specifically trained.’ 7 Samuel R. Bowman, ’Eight
Things to Know about Large Language Models,’ ArXiv Preprint, Apr. 2, 2023. 8
Proposed EU Artificial Intelligence Act, Article 43.1b. 9 National Institute of
Standards and Technology, ‘Artificial Intelligence Risk Management Framework
(AI RMF 1.0),’ U.S. Department of Commerce, Jan. 2023. 10 International
standards bodies such as IEC, ISO and ITU can also help in developing standards
that address risks from advanced AI systems, as they have highlighted in
response to FLI’s call for a pause. 11 See, e.g., the Holistic Evaluation of
Language Models approach by the Center for Research on Foundation Models: Rishi
Bommassani, Percy Liang, & Tony Lee, ‘Language Models are Changing AI: The
Need for Holistic Evaluation’. 12 OpenAI described weaponization risks of GPT-4
on p.12 of the “GPT-4 System Card.” 13 See, e.g., the following benchmark for
assessing adverse behaviors including power-seeking, disutility, and ethical
violations: Alexander Pan, et al., ‘Do the Rewards Justify the Means? Measuring
Trade-os Between Rewards and Ethical Behavior in the MACHIAVELLI Benchmark,’
ArXiv Preprint, Apr. 6, 2023. FUTURE OF LIFE INSTITUTE 7 residual risks deemed
tolerable are disclosed and are subject to established protocols for minimizing
harm. 2. Regulate organizations’ access to computational power At present, the
most advanced AI systems are developed through training that requires an
enormous amount of computational power - ‘compute’ for short. The amount of
compute used to train a general-purpose system largely correlates with its
capabilities, as well as the magnitude of its risks. Today’s most advanced
models, like OpenAI’s GPT-4 or Google’s PaLM, can only be trained with
thousands of specialized chips running over a period of months. While chip
innovation and better algorithms will reduce the resources required in the
future, training the most powerful AI systems will likely remain prohibitively
expensive to all but the best-resourced players. Figure 1. OpenAI is estimated
to have used approximately 700% more compute to train GPT-4 than the next
closest model (Minerva, DeepMind), and 7,000% more compute than to train GPT-3
(Davinci). Depicted above is an estimate of compute used to train GPT-4
calculated by Ben Cottier at Epoch, as oicial training compute details for
GPT-4 have not been released. Data from: Sevilla et al., ‘Parameter, Compute
and Data Trends in Machine Learning,’ 2021 [upd. Apr. 1, 2023]. In practical
terms, compute is more easily monitored and governed than other AI inputs, such
as talent, data, or algorithms. It can be measured relatively easily and the
supply chain for advanced AI systems is highly centralized, which means
governments can leverage such FUTURE OF LIFE INSTITUTE 8 measures in order to
limit the harms of large-scale models.14 To prevent reckless training of the
highest risk models, we recommend that governments make access to large amounts
of specialized computational power for AI conditional upon the completion of a
comprehensive risk assessment. The risk assessment should include a detailed
plan for minimizing risks to individuals, communities, and society, consider
downstream risks in the value chain, and ensure that the AI labs conduct
diligent know-yourcustomer checks. Successful implementation of this
recommendation will require governments to monitor the use of compute at data
centers within their respective jurisdictions.15 The supply chains for AI chips
and other key components for high-performance computing will also need to be
regulated such that chip firmware can alert regulators to unauthorized large
training runs of advanced AI systems.16 In 2022, the U.S. Department of
Commerce’s Bureau of Industry and Security instituted licensing requirements17
for export of many of these components in an eort to monitor and control their
global distribution. However, licensing is only required when exporting to
certain destinations, limiting the capacity to monitor aggregation of equipment
for unauthorized large training runs within the United States and outside the
scope of export restrictions. Companies within the specified destinations have
also successfully skirted monitoring by training AI systems using compute
leased from cloud providers.18 We recommend expansion of know-your-customer
requirements to all high-volume suppliers for high-performance computing
components, as well as providers that permit access to large amounts cloud
compute. 3. Establish capable AI agencies at national level AI is developing at
a breakneck pace and governments need to catch up. The establishment of AI
regulatory agencies helps to consolidate expertise and reduces the risk of a
patchwork approach. The UK has already established an Oice for Artificial
Intelligence and the EU is currently legislating for an AI Board. Similarly, in
the US, Representative Ted Lieu has announced legislation to create a
non-partisan AI Commission with the aim of establishing a regulatory agency.
These eorts need to be sped up, taken up around the world and, eventually,
coordinated within a dedicated international body. 14 Jess Whittlestone et al.,
‘Future of compute review - submission of evidence’, Aug. 8, 2022. 15 Please
see fn. 14 for a detailed proposal for government compute monitoring as drafted
by the Centre for Long-Term Resilience and several sta members of AI lab
Anthropic. 16 Yonadav Shavit at Harvard University has proposed a detailed system
for how governments can place limits on how and when AI systems get trained. 17
Bureau of Industry and Security, Department of Commerce, ‘Implementation of
Additional Export Controls: Certain Advanced Computing and Semiconductor
Manufacturing Items; Supercomputer and Semiconductor End Use; Entity List
Modification‘, Federal Register, Oct. 14, 2022. 18 Eleanor Olcott, Qianer Liu,
& Demetri Sevastopulo, ‘Chinese AI groups use cloud services to evade US
chip export control,’ Financial Times, Mar. 9, 2023. FUTURE OF LIFE INSTITUTE 9
We recommend that national AI agencies be established in line with a
blueprint19 developed by Anton Korinek at Brookings. Korinek proposes that an
AI agency have the power to: • Monitor public developments in AI progress and define
a threshold for which types of advanced AI systems fall under the regulatory
oversight of the agency (e.g. systems above a certain level of compute or that
aect a particularly large group of people). • Mandate impact assessments of AI
systems on various stakeholders, define reporting requirements for advanced AI
companies and audit the impact on people’s rights, wellbeing, and society at
large. For example, in systems used for biomedical research, auditors would be
asked to evaluate the potential for these systems to create new pathogens. •
Establish enforcement authority to act upon risks identified in impact
assessments and to prevent abuse of AI systems. • Publish generalized lessons
from the impact assessments such that consumers, workers and other AI
developers know what problems to look out for. This transparency will also
allow academics to study trends and propose solutions to common problems.
Beyond this blueprint, we also recommend that national agencies around the
world mandate record-keeping of AI safety incidents, such as when a facial
recognition system causes the arrest of an innocent person. Examples include
the non-profit AI Incident Database and the forthcoming EU AI Database created
under the European AI Act.20 4. Establish liability for AI-caused harm AI
systems present a unique challenge in assigning liability. In contrast to
typical commercial products or traditional software, AI systems can perform in
ways that are not well understood by their developers, can learn and adapt after
they are sold and are likely to be applied in unforeseen contexts. The ability
for AI systems to interact with and learn from other AI systems is expected to
expedite the emergence of unanticipated behaviors and capabilities, especially
as the AI ecosystem becomes more expansive and interconnected. Several plug-ins
have already been developed that allow AI systems like ChatGPT to perform tasks
through other online services (e.g. ordering food delivery, booking travel,
making reservations), broadening the range of potential real-world harms that
can result from their use and further complicating the assignment of
liability.21 OpenAI’s GPT-4 system card references an instance of the system
explicitly deceiving a human into bypassing a CAPTCHA botdetection system using
TaskRabbit, a service for soliciting freelance labor.22 When such systems make
consequential decisions or perform tasks that cause harm, assigning
responsibility for that harm is a complex legal challenge. Is the harmful
decision the fault of 19 Anton Korinek, ‘Why we need a new agency to regulate
advanced artificial intelligence: Lessons on AI control from the Facebook
Files,’ Brookings, Dec. 8 2021. 20 Proposed EU Artificial Intelligence Act,
Article 60. 21 Will Knight & Khari Johnson, ‘Now That ChatGPT is Plugged
In, Things Could Get Weird,’ Wired, Mar. 28, 2023. 22 OpenAI, ‘GPT-4 System
Card,’ Mar. 23, 2023, p.15. FUTURE OF LIFE INSTITUTE 10 the AI developer,
deployer, owner, end-user, or the AI system itself? Key among measures to
better incentivize responsible AI development is a coherent liability framework
that allows those who develop and deploy these systems to be held responsible
for resulting harms. Such a proposal should impose a financial cost for failing
to exercise necessary diligence in identifying and mitigating risks, shifting
profit incentives away from reckless empowerment of poorly-understood systems
toward emphasizing the safety and wellbeing of individuals, communities, and
society as a whole. To provide the necessary financial incentives for
profit-driven AI developers to exercise abundant caution, we recommend the
urgent adoption of a framework for liability for AIderived harms. At a minimum,
this framework should hold developers of general-purpose AI systems and AI
systems likely to be deployed for critical functions23 strictly liable for
resulting harms to individuals, property, communities, and society. It should
also allow for joint and several liability for developers and downstream
deployers when deployment of an AI system that was explicitly or implicitly
authorized by the developer results in harm. 5. Introduce measures to prevent
and track AI model leaks Commercial actors may not have suicient incentives to
protect their models, and their cyberdefense measures can often be insuicient.
In early March 2023, Meta demonstrated that this is not a theoretical concern,
when their model known as LLaMa was leaked to the internet.24 As of the date of
this publication, Meta has been unable to determine who leaked the model. This lab
leak allowed anyone to copy the model and represented the first time that a
major tech firm’s restricted-access large language model was released to the
public. Watermarking of AI models provides eective protection against
stealing, illegitimate redistribution and unauthorized application, because
this practice enables legal action against identifiable leakers. Many digital
media are already protected by watermarking - for example through the embedding
of company logos in images or videos. A similar process25 can be applied to
advanced AI models, either by inserting information directly into the model
parameters or by training it on specific trigger data. We recommend that
governments mandate watermarking for AI models, which will make it easier for AI
developers to take action against illegitimate distribution. 6. Expand
technical AI safety research funding The private sector under-invests in
research that ensures that AI systems are safe and secure. Despite nearly USD
100 billion of private investment in AI in 2022 alone, it is estimated that
only about 100 full-time researchers worldwide are specifically working to
ensure AI is safe 23 I.e., functions that could materially aect the wellbeing
or rights of individuals, communities, or society. 24 Joseph Cox, ‘Facebook’s
Powerful Large Language Model Leaks Online,’ VICE, Mar. 7, 2023. 25 For a
systematic overview of how watermarking can be applied to AI models, see:
Franziska Boenisch, ‘A Systematic Review on Model Watermarking of Neural
Networks,’ Front. Big Data, Sec. Cybersecurity & Privacy, Vol. 4, Nov. 29,
2021. FUTURE OF LIFE INSTITUTE 11 and properly aligned with human values and
intentions.26 In recent months, companies developing the most powerful AI
systems have either downsized or entirely abolished their respective
‘responsible AI’ teams.27 While this partly reflects a broader trend of mass
layos across the technology sector, it nonetheless reveals the relative
deprioritization of safety and ethics considerations in the race to put new
systems on the market. Governments have also invested in AI safety and ethics
research, but these investments have primarily focused on narrow applications
rather than on the impact of more general AI systems like those that have
recently been released by the private sector. The US National Science
Foundation (NSF), for example, has established ‘AI Research Institutes’ across
a broad range of disciplines. However, none of these institutes are
specifically working on the large-scale, societal, or aggregate risks presented
by powerful AI systems. To ensure that our capacity to control AI systems keeps
pace with the growing risk that they pose, we recommend a significant increase
in public funding for technical AI safety research in the following research
domains: • Alignment: development of technical mechanisms for ensuring AI
systems learn and perform in accordance with intended expectations, intentions,
and values. • Robustness and assurance: design features to ensure that AI
systems responsible for critical functions28 can perform reliably in unexpected
circumstances, and that their performance can be evaluated by their operators.
• Explainability and interpretability: develop mechanisms for opaque models to
report the internal logic used to produce output or make decisions in
understandable ways. More explainable and interpretable AI systems facilitate
better evaluations of whether output can be trusted. In the past few months,
experts such as the former Special Advisor to the UK Prime Minister on Science
and Technology James W. Phillips29 and a Congressionally-established US
taskforce have called for the creation of national AI labs as ‘a shared
research infrastructure that would provide AI researchers and students with
significantly expanded access to computational resources, high-quality data,
educational tools, and user support.’30 Should governments move forward with
this concept, we propose that at least 25% of resources made available through
these labs be explicitly allocated to technical AI safety projects. 26 This
figure, drawn from , ‘The AI Arms Race is Changing Everything,’ (Andrew R. Chow
& Billy Perrigo, TIME, Feb. 16, 2023 [upd. Feb. 17, 2023]), likely
represents a lower bound for the estimated number of AI safety researchers.
This resource posits a significantly higher number of workers in the AI safety
space, but includes in its estimate all workers ailiated with organizations
that engage in AI safety-related activities. Even if a worker has no
involvement with an organization’s AI safety work or research eorts in
general, they may still be included in the latter estimate. 27 Christine
Criddle & Madhumita Murgia, ‘Big tech companies cut AI ethics sta, raising
safety concerns,’ Financial Times, Mar. 29, 2023. 28 See fn. 23, supra. 29
Original call for a UK government AI lab is set out in this article. 30 For the
taskforce’s detailed recommendations, see: ‘Strengthening and Democratizing the
U.S. Artificial Intelligence Innovation Ecosystem: An Implementation Plan for a
National Artificial Intelligence Research Resource,’ National Artificial
Intelligence Research Resource Task Force Final Report, Jan. 2023. FUTURE OF
LIFE INSTITUTE 12 7. Develop standards for identifying and managing
AI-generated content and recommendations The need to distinguish real from
synthetic media and factual content from ‘hallucinations’ is essential for
maintaining the shared factual foundations underpinning social cohesion.
Advances in generative AI have made it more diicult to distinguish between
AI-generated media and real images, audio, and video recordings. Already we
have seen AI-generated voice technology used in financial scams.31 Creators of
the most powerful AI systems have acknowledged that these systems can produce
convincing textual responses that rely on completely fabricated or
out-of-context information.32 For society to absorb these new technologies, we
will need eective tools that allow the public to evaluate the authenticity and
veracity of the content they consume. We recommend increased funding for
research into techniques, and development of standards, for digital content
provenance. This research, and its associated standards, should ensure that a
reasonable person can determine whether content published online is of
synthetic or natural origin, and whether the content has been digitally
modified, in a manner that protects the privacy and expressive rights of its
creator. We also recommend the expansion of ‘bot-or-not’ laws that require
disclosure when a person is interacting with a chatbot. These laws help prevent
users from being deceived or manipulated by AI systems impersonating humans,
and facilitate contextualizing the source of the information. The draft EU AI
Act requires that AI systems be designed such that users are informed they are
interacting with an AI system,33 and the US State of California enacted a
similar bot disclosure law in 2019.34 Almost all of the world’s nations,
through the adoption of a UNESCO agreement on the ethics of AI, have
recognized35 ‘the right of users to easily identify whether they are
interacting with a living being, or with an AI system imitating human or animal
characteristics.’ We recommend that all governments convert this agreement into
hard law to avoid fraudulent representations of natural personhood by AI from outside
regulated jurisdictions. Even if a user knows they are interacting with an AI
system, they may not know when that system is prioritizing the interests of the
developer or deployer over the user. These systems may appear to be acting in
the user’s interest, but could be designed or employed to serve other
functions. For instance, the developer of a general-purpose AI system could be
financially incentivized to design the system such that when asked about a
product, it preferentially recommends a certain brand, when asked to book a
flight, it subtly prefers a certain airline, when asked for news, it provides
only media advocating specific viewpoints, and when asked for medical advice,
it prioritizes diagnoses that are treated with more profitable pharmaceutical
31 Pranshu Verma, ‘They thought loved ones were calling for help. It was an AI
scam.’ The Washington Post, Mar. 5, 2023. 32 Tiany Hsu & Stuart A.
Thompson, ‘Disinformation Researchers Raise Alarms About A.I. Chatbots,’ The
New York Times, Feb. 8, 2023 [upd. Feb. 13, 2023]. 33 Proposed EU Artificial
Intelligence Act, Article 52. 34 SB 1001 (Hertzberg, Ch. 892, Stats. 2018). 35
Recommendation 125, ‘Outcome document: first draft of the Recommendation on the
Ethics of Artificial Intelligence,’ UNESCO, Sep. 7, 2020, p. 21. FUTURE OF LIFE
INSTITUTE 13 drugs. These preferences could in many cases come at the expense
of the end user’s mental, physical, or financial well-being. Many jurisdictions
require that sponsored content be clearly labeled, but because the provenance
of output from complex general-purpose AI systems is remarkably opaque, these
laws may not apply. We therefore recommend, at a minimum, that
conflict-of-interest trade-os should be clearly communicated to end users
along with any aected output; ideally, laws and industry standards should be
implemented that require AI systems to be designed and deployed with a duty to
prioritize the best interests of the end user. Finally, we recommend the
establishment of laws and industry standards clarifying and the fulfillment of
‘duty of loyalty’ and ‘duty of care’ when AI is used in the place of or in
assistance to a human fiduciary. In some circumstances – for instance,
financial advice and legal counsel – human actors are legally obligated to act
in the best interest of their clients and to exercise due care to minimize
harmful outcomes. AI systems are increasingly being deployed to advise on these
types of decisions or to make them (e.g. trading stocks) independent of human
input. Laws and standards towards this end should require that if an AI system
is to contribute to the decision-making of a fiduciary, the fiduciary must be
able to demonstrate beyond a reasonable doubt that the AI system will observe
duties of loyalty and care comparable to their human counterparts. Otherwise,
any breach of these fiduciary responsibilities should be attributed to the
human fidiciary employing the AI system. FUTURE OF LIFE INSTITUTE 14 Conclusion
The new generation of advanced AI systems is unique in that it presents
significant, welldocumented risks, but can also manifest high-risk capabilities
and biases that are not immediately apparent. In other words, these systems may
perform in ways that their developers had not anticipated or malfunction when
placed in a dierent context. Without appropriate safeguards, these risks are
likely to result in substantial harm, in both the near- and longerterm, to
individuals, communities, and society. Historically, governments have taken
critical action to mitigate risks when confronted with emerging technology
that, if mismanaged, could cause significant harm. Nations around the world
have employed both hard regulation and international consensus to ban the use
and development of biological weapons, pause human genetic engineering, and
establish robust government oversight for introducing new drugs to the market.
All of these eorts required swift action to slow the pace of development, at
least temporarily, and to create institutions that could realize eective
governance appropriate to the technology. Humankind is much safer as a result.
We believe that approaches to advancement in AI R&D that preserve safety
and benefit society are possible, but require decisive, immediate action by
policymakers, lest the pace of technological evolution exceed the pace of
cautious oversight. A pause in development at the frontiers of AI is necessary
to mobilize the instruments of public policy toward commonsense risk
mitigation. We acknowledge that the recommendations in this brief may not be fully
achievable within a six month window, but such a pause would hold the moving
target still and allow policymakers time to implement the foundations of good
AI governance. The path forward will require coordinated eorts by civil
society, governments, academia, industry, and the public. If this can be
achieved, we envision a flourishing future where responsibly developed AI can
be utilized for the good of all humanity
https://futureoflife.org/wp-content/uploads/2023/04/FLI_Policymaking_In_The_Pause.pdf
Hello, an amazing Information dude. Thanks for sharing this nice information with us. NMN Supplement Australia
AtbildētDzēst