Singularity Summit 2009
Introduction
Michael Vassar, Singularity Institute
Shaping the intelligence explosion
Anna Salamon, Singularity Institute
Technical roadmap for whole brain emulation
Anders Sandberg, Future of Humanity Institute
The time is now: As a species and as individuals we need whole brain emulation
Randal Koene, Fatronik-Tecnalia Foundation
Technological convergence leading to artificial general intelligence
Itamar Arel, University of Tennessee
Pathways to beneficial artificial general intelligence: Virtual pets, robot
children, artificial bioscientists, and beyond
Ben Goertzel, Novamente
Neural substrates of consciousness and the 'conscious pilot' model
Stuart Hameroff, University of Arizona
Quantum computing: What it is, what it is not, what we have yet to learn
Michael Nielsen
DNA: Not merely the secret of life
Ned Seeman, New York University
Compression progress: The algorithmic principle behind curiosity, creativity,
art, science, music, humor
Juergen Schmidhuber, IDSIA
Conversation on the Singularity
Stephen Wolfram and Gregory Benford
Simulation and the Singularity
David Chalmers, Australian National University
Choice machines, causality, and cooperation
Gary Drescher
Synthetic neurobiology: Optically engineering the brain to augment its function
Ed Boyden, MIT Media Lab
Foundations of intelligent agents
Marcus Hutter, Australian National University
Cognitive ability: Past and future enhancements and implications
William Dickens, Northeastern University
The ubiquity and predictability of the exponential growth of information
technology
Ray Kurzweil, Kurzweil Technologies
More than Moore: Comparing forecasts of technological progress
Bela Nagy, Santa Fe Institute
The "petaflop macroscope"
Gary Wolf, Wired Magazine
Collaborative networks in scientific discovery
Michael Nielsen
How does society identify experts and when does it work?
Robin Hanson, George Mason University
Artificial biological selection for longevity
Gregory Benford, University of California, Irvine
Critics of the Singularity
Ray Kurzweil, Kurzweil Technologies
The finger of AI: Automated electrical vehicles and oil independence
Brad Templeton, Electronic Frontier Foundation
The fallibility and improvability of the human mind
Gary Marcus, New York University
Macroeconomics and Singularity
Peter Thiel, Clarium Capital Management
The Singularity and the Methuselarity: Similarities and differences
Aubrey De Grey, SENS Foundation
Cognitive biases and giant risks
Eliezer Yudkowsky, Singularity Institute
How much it matters to know what matters: A back of the envelope calculation
Anna Salamon, Singularity Institute
Summit Review
By David Orban
Singularity Hub, October 5, 2009
Edited by Andy Ross
The Singularity Summit 2009, New York, October 3-4, was a resounding success. Over 800 attendees crowded the venue at the 92 Street Y, and there were over 30 speeches and panels in a row.
From the creation of an artificial general intelligence, the whole brain
emulation approach, synthetic neurobiology, and intelligent agents, there were
many technical sessions, interspersed with others on cognition and its biases,
quantum computing, or subjects such as autonomous vehicles and the petaflop
macroscope.
A
welcome addition to the program was a conversation between science-fiction
author, and now chairman of Genescient, Gregory Benford, and the developer of
symbolic computation system Mathematica, Stephen Wolfram.
The concepts of artificial general intelligence, of an intelligence explosion,
and of the technological singularity are becoming more and more widely accepted.
But the bounds for these achievements or events do not necessarily
converge. Ray Kurzweil persists in his precise projections of fundamental
milestones in the future.
An almost universal need appeared to find a better and stronger connection
between the ideas, theories, and models exposed at the conference and their full
implementation.
Eliezer Yudkowsky illustrated the unacceptable downsides of not
properly dealing with the issues of the singularity and of artificial general
intelligence in his talk on cognitive biases and giant risks.
Singularity Summit 2009
By Stuart Fox
Popular Science, October 3-4, 2009
Edited by Andy Ross
Blog 1 Open The Pod Bay Door, HAL
Ray Kurzweil's concept of the Singularity rests on two axioms: that computers
will become more intelligent than humans, and that humans and computers will
merge, allowing us access to that increased thinking power.
According to Anna Salamon, artificial intelligence greater than our own is inevitable and dangerous. She argued that biological brains have finite intellectual capacity.
Salamon believes we will create super computers to solve those problems for us.
She worries that if humans and AI have divergent goals, we could find ourselves
in competition with the AI for resources to achieve those different goals. Salamon advocates starting now to ensure that
human-assisting missions get hardwired into the basic architecture of
artificial intelligence.
Anders Sandberg believes that engineers have to base their first
attempts at AI on the human
brain. So the first artificial brain would contain elements of the personality
of the test subject that the artificial brain copied. These traits could become
locked into all artificial intelligence as the initial AI software proliferates.
Blog 2 Just How's This Thing Gonna Work, Anyways?
How are we going to create artificial intelligence, and how are we going to
integrate ourselves with this advanced technology? Luckily, philosopher David
Chalmers was there to break it all down.
Chalmers rejected the idea of brain emulation as the path to super-intelligent
AI. He thinks that we have to evolve AI by planting
computer programs in a simulated environment and selecting the smartest bots.
Basically, set up the Matrix, but with only artificial inhabitants.
To ensure that the resultant AI adheres to pro-human values, we would have to
set up a "leak-proof" world where we control what goes in, and can prevent any
of the artificial consciousness from becoming aware of us too early and
escaping.
As Chalmers sees it, the second the artificial personalities become as smart as
us, they will emulate our leap by creating AI even smarter than themselves
inside their simulated world. Essentially, they will undergo their own, digital
Singularity.
This will start a chain reaction that quickly leads to a digital intelligence
far greater than anything we ever imagined. Unless, of course, the first AI more
intelligent than us uses that additional foresight to realize creating
intelligence greater than itself is a bad idea, and cuts off the entire process.
But assuming that AI does manage to get smarter than us, we will have to either
integrate with it, coexist with it as a lower form of life, or pull the plug.
Chalmers sees integration as the only way to go. He advocates physically
replacing one neuron at a time with a digital equivalent, while the person is
awake, so as to retain continuity of personality.
Blog 3 Supreme Mathematics of Gods and Earths
Stephen Wolfram believes that fundamental programs underlie all the behavior in our
universe, as well as many phenomena that don't exist in a universe with our
physics. These computations that Wolfram identifies as embedded in
reality exist independently of our observation.
Wolfram calls the total set of all possible programs the computational universe. By
running mathematical experiments, examining the natural world and decoding the
behavior of reality, we can explore this universe,
uncovering programs new to humanity, but not new to the universe.
Wolfram likened these programs to minerals like crystals and magnetic metals.
He described a world where scientists and mathematicians mine the
computational universe for new programs.
Wolfram identifies computerized computational mining as the
catalyst for the emergence of artificial intelligence. But he isn't concerned
that this AI will immediately threaten our extinction. After all, the program
only exists to find new knowledge. How could killing us help in that goal?
Not everyone at the conference bought this idea of a benign artificial
intelligence. Maybe this pervasive fear of AI-led extermination just reflects
our own inability to imagine a consciousness without the aggressive need to
destroy humanity.
Blog 4 Thus Spake Kurzweil
Kurzweil is the man everyone came to see. After the standing ovation died down,
the auditorium reached its quietest point yet, as the collected skeptics,
crazies, and disciples waited to hear from the first prophet of Singularity.
Kurzweil shored up the faithful, calming any doubts they had about the
sheer ambition of his claims, and presenting even stronger evidence that the
Singularity is inevitable and impending. The blatant clarity and simplicity of
his argument and evidence left no doubts about Kurzweil's profound intellect.
Since 1890, computing power has become a trillion times more powerful, and a
billion times more powerful in the last 25 years. A single computer will equal
the storage capacity and speed of the human brain by around 2029. And once a
computer can map out every single neuron, connection, and firing of a brain,
someone will make a digital version.
Kurzweil didn't convince me that a digital brain will spontaneously assume
human-like consciousness and self-awareness. In fact, he didn't convince me that
anyone had even the slightest clue as to what will really happen once we cross
that threshold.
I was reminded of the Human Genome Project, which assumed that once every gene
got mapped out, it would be easy to put a person together from scratch. Now the
simple theory that DNA codes, RNA prints, and protein acts seems increasingly
simplistic and naive.
Neuroscience will soon start revealing similar complexities in the brain. And as
the process of consciousness proves more and more intricate, the computing power
needed to reproduce it will rise and rise, pushing back the date of the
Singularity.
Kurzweil's presentation left me with no doubt that he was on to something. But
he also left me with a strong belief that no knows what that something is, or
when it will really be here.
Blog 5 Ten Unanswered Questions For Our Future Robot Overlords
10. Is there just one kind of consciousness or intelligence?
Assuming there are different forms of intelligence, how do we know machines
won't take on a new one that we won't recognize as intelligence? And if there
are different kinds of intelligence, are there different kinds of consciousness,
too? Could a machine arrive at a new kind of consciousness that we don't
recognize, leading us to miss the Singularity?
9. How will you use your digital intelligence to kill us all?
A lot of people spent the conference worrying about our eventual extinction at
the hands of our automaton creations. But for all that paranoia, no one really
explained how a computer program could manage to kill me. Will it hack into the
nuclear missile command and launch all the nukes? Will it crash all the planes?
And couldn't we just pull the plug?
8. Are you Tommy: deaf, dumb, and blind?
When the first artificial brain comes online, how can its first thought be
anything other than "holy crap, I'm blind!" A disembodied intelligence in a
machine will exist with a serious lack of senses. Maybe it can see and hear, but
feel? Doubtful. How does a consciousness that can't feel keep from freaking out?
7. Do you have emotions?
Can AI become depressed? The first one will no doubt be rather lonely. How will
being the first (and only) member of a species affect the AI's development and
relationships? The first digital consciousness may come into the world without
anyone who can sympathize. Not really the kind of being I want with access to
all our weapons and economic tools.
6. Are humans more similar to your AI construct than we thought?
Just how computer-like is our brain already? Ours brains already run software,
of sorts, that result in biologically similar brains producing vastly different
personalities. Is it possible the Singularity will occur not because we create
machines that resemble the human brain, but because we uncover just how
computer-like the human brain is naturally?
5. How much does programming influence your free will?
In the discussions about avoiding a robo-apocalypse, speaker after speaker
stressed the need to teach digital consciousnesses to have human values. And
many people wondered why we couldn't just program the robots not to kill us.
Well, presumably we would, but once the computer programs achieve self-awareness
and free will, couldn't they choose not to follow that programing?
4. Do you have a subconscious?
If AI minds are as complex as human brains, does that mean they will have areas
that they cannot understand, control, or access? Can AI have irrational beliefs
or psychological problems? If the AI thinks we're their god, or at the very
least their creator, could it have an oedipal problem? If so, that might explain
why it tries to kill us.
3. Will you help us transcend the less pleasant aspects of being human?
In addition to granting immortality and making everyone well-nigh omniscient,
won't the Singularity also provide the ultimate avenue for people to disseminate
the lust, greed, and hatred humans have pursued for tens of thousands of years?
Forget about the AI killing us, I'm still worried about the other humans.
2. Do you care about anything at all?
What's to say that an intelligence vastly greater than our own won't uncover the
pointlessness of life, become a nihilist, and turn itself off? Or, what if it
simply doesn't care about humans? Everyone at the conference predicted a very
needy AI, but no one could answer why the AI wouldn't be just as likely to
withdraw from humanity as engage it.
1. What if someone threw a Singularity and no one came?
If the Singularity only affects one small group of humans, while the rest either
can't afford it or simply don't care to participate, what happens to the
transhumanist future?
What Does a Beer Taste Like After the Singularity?
By
Glenn Derene
Popular Mechanics, October 5, 2009
Edited by Andy Ross
Imagine a techno-futurist rapture when artificial intelligence becomes smarter than human
intelligence, and computers are able to improve and refine themselves at an
accelerated pace. Our machines will make machines that are even smarter than
them, then those machines will make other machines that are smarter than the
previous machines, and so on, making the conventional human mind more and more
of an anachronism.
Most pop-culture visions of this sort of future conclude
that any machine that achieved superior intelligence would naturally decide
that humans and humanity in general were no longer necessary, except perhaps as biofuel.
That's why many singularians support the idea of "uploading" one's
consciousness to the machines once the flesh and blood support system of
intelligence becomes essentially irrelevant. With your mind freed into the
digital realm, you will be immortal.
Gary Marcus, director of the NYU Center for Child Language,
described the human mind as an evolutionary kluge, producing a wonderfully
refined sense of vision, but a terribly deficient cue-based memory recall
system. Altering the mind with a more logical computer system could reduce
errors and improve our basic capacity to reason.
If it were possible to improve your memory with a digital device, then everybody
would want one, because not having such a device would put you at a
disadvantage. Then an escalation of biodigital enhancement would naturally occur
until some people were walking around with more microchips than neurons. At some
point the hand-off between human intelligence and machine intelligence would
have occurred. And that's just one possible singularity scenario.
But I think this stuff is a lot harder than these folks make it out to be.
Before we can decide what we want from artificial intelligence, we need to
figure out just what we mean when we describe human intelligence.
Even if we accepted that it was possible to digitize the broad, ever-evolving
spectrum that is human intelligence, add your own consciousness to it, and then
accelerate the heck out of it and spin the whole concoction indefinitely into
the future, what would the point be, exactly? Once divorced from a corporeal
self, any mind that a person might upload would probably quickly lose anything
resembling a human psychology. A "mind" living in a computer would hardly have
need for food, and so eating a deliciously greasy bacon cheeseburger along with
a cold beer would be unnecessary.
If any computer becomes self-aware enough to start
refining its own intelligence at an accelerated rate, I'm going to unplug that
thing and take a baseball bat to it.
Will Our Robot Overlords Be Friendly?
Ronald Bailey, October 6, 2009
The Singularity Institute for Artificial Intelligence (SIAI) was created to
address the urgent problem of how to create super-smart AIs that are friendly to
human beings. SI fellow Anna Salamon explained in her opening presentation that
smarter intelligences might choose to get rid of us because our matter is not
optimally arranged to achieve their goals.
One way to wind up with human-friendly AIs is to build them on top of uploaded
human brains. Anders Sandberg offered a technical roadmap for whole brain
emulation. "If artificial intelligence does not get here first, we're going to
end up with uploaded humans," maintained Sandberg. He argued that it is possible
to foresee how developments in neuroscience and computer science are leading to
emulation of specific people's brains. Sandberg believes that emulating a human
brain is only 20 years away. But we don't know if a one-to-one emulation would
produce a mind or not. An uploaded mind would no longer be constrained by the
speed at which organic brains can process information.
Randal Koene argued that the time is now to go after mind uploading. Radically
increasing human longevity solves a few problems, but doesn't deal with our
dependence on changeable environments and scarce resources. Nor does it deal
with our intellectual limitations, death by traumatic accidents, or disease.
Koene concludes that we must free the mind from its single fragile substrate.
How might you copy a brain? Perhaps one could use a more advanced version of the
method currently used to map the mouse brain. This is destructive scanning. The
old organic brain is sliced up to produce the emulation. To solve the problem of
maintaining a sense of continuity, Koene suggested using molecular
nanotechnology to replace parts of the brain bit by bit over time.
David Chalmers argued that personal identity would be maintained if the
functional organization of the upload was the same as the original. In addition,
gradual uploading might also be a way to maintain personal identity. Chalmers
also speculated about reconstructive uploading in which a super- smart AI would
scour the world for information about me, then instantiate that information in
the appropriate substrate. Is it me? On the optimistic view, being reconstructed
from the informational debris you left behind would be like waking up from a
long nap.
Ray Kurzweil envisions an intimate integration between humans and their neural
prosthetics. Over time, more and more of the neural processing that makes us who
we are will be located outside our bodies and brains so that "uploading" will
take place gradually. Our uploaded minds will function much faster and more
precisely than our "meat" minds do today. We will join the singularity as our
artificial parts become ascendant.
Chalmers suggested that perhaps we could launch a super-smart self-improving AI
inside a leak-proof computing environment and see how it evolves. If it turns
out to be nice, then we let it out and the singularity takes off and we're all
happy. Kurzweil objected that a leak-proof singularity is impossible. To
determine whether or not the AI was friendly we would have to look into its
environment, and then it would see us and trick us into letting it out. In other
words, it could pretend to be friendly and then zap us once it's present in our
world.
So why is everyone so excited about the singularity? Peter Thiel asked us to
vote on which of seven scenarios we're most worried about:
— Singularity happens and robots kill us all (the Skynet scenario)
— Biotech terrorism using something more virulent than smallpox and Ebola
combined
— Nanotech grey goo escapes and eats up all organic matter
— Israel and Iran engage in thermonuclear war that goes global
— A one-world totalitarian state arises
— Runaway global warming
— The singularity takes too long to happen
The last one is what worries Thiel. Without rapid technological progress,
economic growth in developed countries like the United States, Western Europe,
and Japan is not going be enough to address looming needs.
If you want to be alive when the singularity gets here, consider Aubrey de
Grey's proposals for anti-aging research. In his view, progress in regenerative
medicine could achieve longevity escape velocity in which researchers develop
anti-aging therapies faster than a person approaches death from aging.
Anna Salamon argued during the summit that an intelligence explosion, or
singularity, can't be controlled once it starts. She likened the situation to
trying to put a leash on an exploding atomic bomb. Creating a super-smart
friendly AI is the "hardest goal humans have ever tried."
A Brief Review
Aaron Diaz, October 7, 2009
I was very privileged to attend the Singularity Summit in New York. Every
speaker treated the subject matter as a serious and legitimate topic of
discussion and investigation.
Ray Kurzweil's talks were very good. I was impressed by the astonishing amount
of evidence, detail, and overall research he puts into his thesis. He really
looks a lot younger than a man of 61.
Eliezer Yudkowsky examined cognitive bias. He showed how even a rational
person's mind can be hacked due to its messy spaghetti code. Symptoms of
cognitive bias include tending to believe more specific or dramatic threats over
more realistic ones. With relation to the Singularity, the Terminator scenario
is popular because it's dramatic.
Peter Thiel believes that the topic that needs to be discussed further is what
happens if the Singularity doesn't happen or doesn't happen fast enough. He
seems to thnk if it doesn't develop quickly enough, there could be an incident
similar to the dot-com crash, only much more devastating.
Aubrey de Grey graphed life expectancy paths among different age groups today.
It was a very thorough talk and involved some of the hardest science out there.
I feel this is one angle of Transhumanism that should get more face time.
Postscript
Al Fin, October 7, 2009
The singularity is only the latest of names for an idea that has been around for
centuries. Ideas of exchanging a mortal body for an immortal one, or exchanging
a limited mind for a virtually unlimited one, have been around for millennia in
religions. The singularity movement can be seen as a religious faith. The
combination of religious faith and goal orientation with scientific and
engineering rigor can lead to something truly amazing.
More on Kurzweil:
My selections from an IEEE Spectrum
Special Issue on the Singularity
Kurzweil says anyone alive in
2040 or so could be close to immortal
AR I should have been there. Too bad — I shall have to rest content with watching the videos and reading the secondary literature. Then I shall publish a devastating reply — as my next book.
