Image: Bryan Christie Design
IEEE SPECIAL REPORT
THE SINGULARITY
Reports below edited by Andy Ross
Waiting for the Rapture
The singularity has been called the rapture of the geeks.
The singularity is supposed to begin shortly after engineers build the first
computer with greater-than-human intelligence. That achievement will trigger a
series of cycles in which superintelligent machines beget even smarter machine
progeny, going from generation to generation in weeks or days rather than
decades or years. This will spark explosive economic growth and a
technoindustrial rampage that will suck us beyond the event horizon.
But the singularity is much more than a sci-fi subgenre. In the coming years, as
computers become stupendously powerful and as electronics and other technologies
begin to enhance and fuse with biology, life really is going to get more
interesting.
So we invited articles from people who impressed us with their achievements and
writings on subjects central to the singularity idea. On consciousness, we have
John Horgan. We also have Christof Koch and Giulio Tononi, neuroscientists who
specialize in consciousness. Rodney Brooks, from MIT, weighs in on the future of
machine intelligence. For the last word, we turned to Vernor Vinge, who launched
the singularity movement in 1993.
That movement has evolved since then into an array of competing hypotheses and
scenarios. But central to them all is the idea of a conscious machine.
Consciousness seems mystical and inextricably linked to organisms. What happens
in the cerebral cortex that turns objective information into subjective
experience? We dont know, but we will someday. No one argues that consciousness
arises from anything but biological processes in the brain. The brain is a
computer.
The Consciousness Conundrum
I would love to believe that we are rapidly approaching the
singularity. Technological singularity comes in many versions, but most involve
bionic brain boosting. At first, we'll become cyborgs. Eventually, we will
abandon our flesh-and-blood selves entirely and upload our digitized psyches
into computers. We will then dwell happily forever in cyberspace.
Intoxicated by the explosive progress of information technologies,
singularitarians foresee a "merger of biological and nonbiological
intelligence," as Ray Kurzweil puts it, that will culminate in "immortal
software-based humans." It will happen not within a millennium, or a century,
but no later than 2030, according to Vinge.
Neuroscientists still do not understand at all how a brain makes a conscious
mind. "No one has the foggiest notion," says the neuroscientist Eric Kandel of
Columbia University Medical Center, in New York City.
Gerald Edelman, a Nobel laureate and director of the Neurosciences Institute, in
San Diego, says singularitarians vastly underestimate the brain's complexity. A
healthy adult brain contains about 100 billion neurons. A single neuron can be
linked via axons (outputs) and dendrites (inputs) across synapses (gaps between
axons and dendrites) to as many as 100 000 other neurons. A typical human brain
has quadrillions of connections among its neurons. Adding to the complexity,
synaptic connections constantly form, strengthen, weaken, and dissolve.
Nevertheless, the brain is a computer. Neurons resemble transistors, processing
electrochemical pulses known as action potentials. With an amplitude of
one-tenth of a volt and a duration of one millisecond, action potentials are
remarkably uniform. Also called spikes, action potentials serve as the brain's
basic units of information.
If the brain contains one quadrillion synapses processing on average 10 action
potentials per second, then the brain performs 10 quadrillion operations per
second. If current trends continue, supercomputers will exceed 10 quadrillion
operations per second within a decade.
Intelligence requires software as much as hardware. In the next couple of
decades, scientists will reverse engineer the brain's software. First, the
brain's programming tricks will be transferred to computers to make them
smarter. Eventually, our personal software will be extracted from our frail
flesh and uploaded into advanced robots or computers.
Neuroscientists suspect that the brain employs a temporal code, in which
information is represented not just in a cell's rate of firing but also in the
precise timing between spikes. Biophysicist William Bialek of Princeton
University calculates that temporal coding would boost the brain's
information-processing capacity close to the Shannon limit.
Edelman has advocated a scheme called neural Darwinism, in which our recognition
of, say, an animal emerges from competition between large populations of neurons
representing different memories. The brain quickly settles on the population
that most closely matches the incoming stimulus.
Wolf Singer of the Max Planck Institute for Brain Research, in Frankfurt, has
won more support for a code involving many neurons firing at the same rate and
time. Singer thinks such synchronous oscillations might play a crucial role in
cognition and perhaps even underpin consciousness.
In 1990, the late Nobel laureate Francis Crick and his colleague Christof Koch
proposed that 40-hertz synchronized oscillations were one of the neuronal
signatures of consciousness. But Singer says the brain probably employs many
different codes in addition to oscillations. He also emphasizes that researchers
are "only at the beginning of understanding" how neural processes "bring forth
higher cognitive and executive functions."
Singer calls the idea of an imminent singularity science fiction. Koch shares
Singer's skepticism. There may be no universal principle governing
neural-information processing, Koch says.
Researchers at the University of Southern California, in Los Angeles, have
designed chips that mimic the firing patterns of tissue in the hippocampus, a
neural structure thought to underpin memory. Biomedical engineering professor
Theodore Berger, a leader of the USC program, has suggested that one day brain
chips might allow us to instantly upload expertise.
Andrew Schwartz, a neural-prosthesis researcher at the University of Pittsburgh,
has shown that monkeys can learn to control robotic arms by means of chips
embedded in the brain's motor cortex. But no one has any idea how memories are
encoded, Schwartz says.
That brings us to the interface problem. For now, electrodes implanted into the
brain remain the only way to precisely observe and fiddle with neurons. It is a
much messier, more difficult, and more dangerous interface than most people
realize. The electrodes must be inserted into the brain through holes drilled in
the skull, posing the risk of infection and brain damage.
Researchers are testing various strategies for improving contact between neurons
and electronics. They are making electrodes out of conducting polymers, coating
electrodes with natural cell-adhesion molecules, and designing arrays of
electrodes that automatically adjust their position.
At Caltech and elsewhere, engineers have designed hollow electrodes that can
inject fluids into the surrounding tissue. The fluids could consist of
nerve-growth factors, neurotransmitters, and other substances. Neuroscientists
are also testing optical devices and genetic switches.
Terry Sejnowski, a neuroscientist at the Salk Institute for Biological Studies,
in San Diego, says the new technologies will make it possible "to selectively
activate and inactivate specific types of neurons and synapses as well as record
from all the neurons in a volume of tissue."
Even singularitarians concede that no existing interface can provide what is
required for bionic convergence and uploading. So they predict that current
interfaces will soon yield to nanobots. They infiltrate the brain, then record
all neural activity and manipulate it by zapping neurons, tinkering with
synaptic links, and so on. The nanobots will be equipped with some sort of Wi-Fi
so that they can communicate with one another as well as with electronic systems
inside and outside the body.
Steven Rose, a neurobiologist at England's Open University, says a lot can be
done to improve the brain's performance through improved drugs, neural
prostheses, and perhaps genetic engineering. But he calls the claims about
imminent consciousness uploading "pretty much crap."
Rose disputes the singularitarians' contention that computers will soon surpass
the brain's computational capacity. He suspects that computation occurs at
scales above and below the level of individual neurons and synapses, via
genetic, hormonal, and other processes. So the brain's total computational power
may be many orders of magnitude greater than what singularitarians profess.
Rose also rejects the basic premise of uploading, that our psyches consist of
nothing more than algorithms that can be transferred from our bodies to entirely
different substrates, whether silicon or glass fibers or quantum computers. The
information processing that constitutes our selves, Rose asserts, evolved within
a social, crafty, emotional, sex-obsessed flesh-and-blood primate.
If the brain were simple enough for us to understand, we wouldn't be smart
enough to understand it.
The singularity is a religious rather than a scientific vision. It is the
rapture for nerds.
Can Machines Be Conscious?
By Christof Koch and Giulio Tononi
In some quarters it is taken for granted that within a
generation, human beings will have an alternative to death: being a ghost in a
machine. You'll be able to upload your mind to a computer. And once you've
reduced your consciousness to patterns of electrons, others will be able to copy
it, edit it, sell it, or pirate it. It might be bundled with other electronic
minds. And, of course, it could be deleted.
Some of the most brilliant minds in human history have pondered consciousness.
We know it arises in the brain, but we don't know how or where in the brain.
Nevertheless, some in the singularity crowd are confident that we are within a
few decades of building a computer that can experience things as we do. It might
be a robot. Or it might just be software a huge, ever-changing cloud of bits
that inhabit an immensely complicated and elaborately constructed virtual
domain.
We are among the few neuroscientists who have devoted a substantial part of
their careers to studying consciousness. Our work has given us a unique
perspective on whether consciousness will ever be artificially created.
We think it will, eventually. But perhaps not in the way that the most popular
scenarios have envisioned it.
Consciousness is part of the natural world. It depends, we believe, only on
mathematics and logic and on the imperfectly known laws of physics, chemistry,
and biology. It does not arise from some magical or otherworldly quality.
In humans and animals, we know that the specific content of any conscious
experience is furnished by parts of the cerebral cortex. If a sector of the
cortex is destroyed, the person will no longer be conscious of whatever aspect
of the world that part of the brain represents. To be conscious also requires
the corticothalamic system to be constantly suffused in a bath of substances
known as neuromodulators, which aid or inhibit the transmission of nerve
impulses.
Much of what goes on in the brain has nothing to do with being conscious.
Widespread damage to the cerebellum, the small structure at the base of the
brain, has no effect on consciousness. Neural activity obviously plays some
essential role in consciousness but in itself is not enough to sustain a
conscious state.
Clinical studies and basic research have given us a complex if still rudimentary
understanding of the myriad processes that give rise to consciousness. We are
still a very long way from being able to use this knowledge to build a conscious
machine. Yet we can list some aspects of consciousness that are not strictly
necessary for building such an artifact.
Consciousness does not seem to require emotions, memory, self-reflection,
language, sensing the world, and acting in it. When we dream, we are virtually
disconnected from the environment but we are conscious, and the corticothalamic
system continues to function more or less as it does in wakefulness.
So although being conscious depends on brain activity, it does not require any
interaction with the environment. Whether the development of consciousness
requires such interactions in early childhood is a different matter.
Being conscious does not require emotion. People who've suffered damage to the
frontal area of the brain may exhibit a flat, emotionless affect. But they still
experience the sights and sounds of the world much the way normal people do.
Primal emotions are useful and perhaps even essential for the survival of a
conscious organism. Likewise, a conscious machine might rely on emotions to make
choices and deal with the complexities of the world. But it could be just a
calculating engine, and yet still be conscious.
Psychologists argue that consciousness requires selective attention. When you
pay attention to something, you become conscious of that thing and its
properties. When your attention shifts, the object fades from consciousness.
Nevertheless, a person can consciously perceive an event or object without
paying attention to it. Conversely, people can attend to events or objects
without consciously perceiving them.
Episodic memory would seem to be an integral part of consciousness. But being
conscious does not require either explicit or working memory.
Self-reflection is another human trait that seems deeply linked to
consciousness. To assess consciousness, psychologists and other scientists often
rely on verbal reports from their subjects. But being conscious does not require
self-reflection. When we become absorbed in some intense perceptual task, we are
vividly conscious of the external world without any need for reflection or
introspection.
Neuroimaging studies suggest that we can be vividly conscious even when the
front of the cerebral cortex, involved in judgment and self-representation, is
relatively inactive. Patients with widespread injury to the front of the brain
demonstrate serious deficits but they appear to have nearly intact perceptual
abilities.
Finally, being conscious does not require language. There are many patients who
lose the ability to understand or use words and yet remain conscious. And
infants, monkeys, dogs, and mice cannot speak.
We assume that a machine does not require anything to be conscious that a
naturally evolved organism doesn't require. A conscious machine does not need to
engage with its environment, or have long-term memory or working memory, or
attention, self-reflection, language, or emotion.
So what are the essential properties of consciousness?
We think the answer has to do with the amount of integrated information that an
organism, or a machine, can generate. Information is classically defined as the
reduction of uncertainty that occurs when one among many possible outcomes is
chosen.
But conscious experience consists of more than just differentiating among many
states. Consider an idealized 1-megapixel digital camera. Even if each
photodiode in the imager were just binary, the number of different patterns that
imager could record is 2 raised to the power 1 million. Yet the camera is
obviously not conscious.
We think that the difference between you and the camera has to do with
integrated information. The 1-megapixel sensor chip isn't a single integrated
system but rather a collection of one million individual, completely independent
photodiodes, each with a repertoire of two states. And a million photodiodes are
collectively no smarter than one photodiode.
By contrast, the repertoire of states available to you cannot be subdivided. You
know this from experience. Underlying this unity is a multitude of causal
interactions among the relevant parts of your brain. And unlike chopping up the
photodiodes in a camera sensor, disconnecting the elements of your brain that
feed into consciousness would have profoundly detrimental effects.
To be conscious,you need to be a single integrated entity with a large
repertoire of states. Your level of consciousness has to do with how much
integrated information you can generate.
The integrated information theory of consciousness, or IIT, is grounded in the
mathematics of information and complexity theory and provides a specific measure
of the amount of integrated information generated by any system comprising
interacting parts. We call that measure F and express it in bits. The larger the
value of F, the larger the entity's conscious repertoire. F is an intrinsic
property of the system, different from the Shannon information that can be sent
through a channel.
IIT suggests a Turing Test for consciousness. According to IIT, consciousness
implies the availability of a large repertoire of states belonging to a single
integrated system. To be useful, those internal states should also be highly
informative about the world. One test would be to ask the machine to describe a
scene in a way that efficiently differentiates the scene's key features from the
immense range of other possible scenes. Humans are fantastically good at this.
So we can test for machine consciousness by showing it a picture and asking it
for a concise description. The machine should be able to extract the gist of the
image and what's happening. The machine should also be able to describe which
objects are in the picture and which are not, as well as the spatial
relationships among the objects and the causal relationships.
No machine or program comes close to pulling off such a feat today. In fact,
image understanding remains one of the great unsolved problems of artificial
intelligence.
To build a conscious machine, we can either copy the mammalian brain or evolve a
machine. Research groups worldwide are already pursuing both strategies.
The Association for the Scientific Study of Consciousness
Executive director: Christof Koch, professor of cognitive and
behavioral biology at Caltech
President-elect: Giulio Tononi, professor of psychiatry at the University of
Wisconsin, Madison
Consciousness as Integrated Information
The integrated-information theory of consciousness, or IIT, is an
attempt to approach consciousness from first principles.
IIT introduces a measure of integrated information, represented by the symbol F
and given in bits, that quantifies the reduction of uncertainty. This measure is
above and beyond the information generated independently within the parts
themselves.
A system with a positive value of F is called a complex. When a complex enters a
particular state of its repertoire, it generates an amount of integrated
information corresponding to F. Thus, a simple photodiode that can detect the
presence or absence of light is a complex with F = 1 bit. The sensor chip of a
digital camera, on the other hand, would not form a complex: as such it would
have F = 0 bits, as each photodiode does its job independently of the others. In
principle, it can be decomposed into individual photodiodes, each with F = 1
bit.
Within the awake human brain, on the other hand, there must be some complex
whose F value is on average very high, corresponding to our large repertoire of
conscious states that are experienced as a whole. Because integrated information
can be generated only within a complex and not outside its boundaries, it
follows that consciousness is subjective and related to a single point of view
or perspective.
Given the vast number of ways even a simple information-processing system can be
decomposed, measuring F can be done only for very simple systems. Also, the
value of F depends on both spatial and temporal scales that determine what
counts as elements and states of a system.
With the aid of computer simulations, one can try out different networks and see
which architectures yield high values of F. Such simulations indicate that high
F requires networks that combine functional specialization with functional
integration so that each element has a unique function within the network and
there are many pathways for interactions among the elements. In very rough
terms, this kind of architecture describes the mammalian thalamocortical system.
The thalamocortical system is the part of the brain that cannot be severely
impaired without loss of consciousness.
Conversely, F is low for systems made up of small, quasi-independent modules.
This suggests that parts of the brain that are organized in an extremely modular
manner, where modules hardly interact, should not contribute directly to
consciousness. The cerebellum offers an example. If the cerebellum is damaged,
consciousness is largely unaffected. Although the cerebellum is a powerful
computer, it is the wrong machine for consciousness, being far too modular to
generate much integrated information.
Computer simulations also indicate that parallel input or output pathways can be
attached to a complex of high F without becoming part of it. This may explain,
for example, why the retina, which is connected to the visual cortex by multiple
parallel pathways, does not directly contribute to visual consciousness.
Simulations show that a complex of high F can be augmented by attaching local
circuits to some of its elements and yet the attached circuits may remain
outside of the high F complex. In the brain, it appears that many computations
that remain unconscious are carried out by cortical and subcortical circuits
that appear to be informationally insulated.
Consciousness is associated with neural architectures that form a a single
entity having a large repertoire of states. It seems that one way to achieve
high values of F is to build a network that is both highly specialized and
highly integrated.
According to this theory, F, and therefore conscious experience, is graded. It
is not an all-or-none property that only sufficiently complex systems possess.
Any physical system with some capacity for integrated information would have
some degree of conscious experience.
Here we have only our first-person evidence to fall back on. Practically
speaking, we can think of the F value associated with dreamless sleep or general
anesthesia as a threshold for consciousness. Anything with a smaller F is no
more conscious than you or I during such a dreamless state.
Do You Need a Quantum Computer to Achieve Machine Consciousness?
Oxford University cosmologist Roger Penrose has surmised that a
yet-to-be-discovered quantum theory of gravity lies at the core of
consciousness. If that is true, you would not be able to upload your
consciousness into a classical machine. It would have to be a machine that
exploited quantum entanglement at the level of its elementary gates. You'd need
a quantum computer, with the processing and memory capacity of a human brain.
However, there is no compelling evidence that brains exploit any of the special
features of quantum mechanics. The components of the nervous system would make
it very difficult to retain entangled states, or qubits, over the necessary
spatial-temporal dimensions. It is likely that to simulate or emulate
brain-based functions, including consciousness, computers built out of
classical, nonquantum gates will suffice.
I Am a Robot
I am a machine. So are you. We are really sophisticated machines
made up of billions and billions of biomolecules that interact according to
rules deriving from physics and chemistry.
If we really are machines and if we learn the rules governing our brains, then
in principle there's no reason why we shouldn't be able to replicate those rules
in, say, silicon and steel. I believe our creation would exhibit genuine
human-level intelligence, emotions, and even consciousness.
One day we will create an artificial general intelligence, or AGI. Some
researchers believe that AGIs will undergo a positive-feedback self-enhancement
until their comprehension of the universe far surpasses our own. Our world,
those individuals say, will change in unfathomable ways after such superhuman
intelligence comes into existence, an event they refer to as the singularity.
Ray Kurzweil and his colleagues believe that this super AGI will be created
either through ever-faster advances in artificial intelligence or by more
biological means, such as "direct brain-computer interfaces, biological
augmentation of the brain, genetic engineering," and "ultrahigh-resolution scans
of the brain followed by computer emulation." They think it will happen sometime
in the next two or three decades.
Some singularitarians believe our world will become a kind of techno-utopia,
with humans downloading their consciousnesses into machines to live a
disembodied, after-death life. Others anticipate a kind of techno-damnation in
which intelligent machines will be in conflict with humans, maybe waging war
against us. Their arguments are plausible, but plausibility is by no means
certainty.
I don't think there is going to be one single sudden technological "big bang"
that springs an AGI into "life." Starting with the mildly intelligent systems we
have today, machines will become gradually more intelligent, generation by
generation. The singularity will be a period, not an event.
This period will encompass a time when we will invent, perfect, and deploy ever
more capable systems, driven by the usual economic and sociological forces.
Eventually, we will create truly artificial intelligences, with cognition and
consciousness recognizably similar to our own. I strongly suspect it won't
happen before 2030.
But I expect the AGIs of the future embodied as robots to emerge gradually
and symbiotically with our society. At the same time, we humans will transform
ourselves. We will incorporate a wide range of advanced sensory devices and
prosthetics to enhance our bodies. As our machines become more like us, we will
become more like them.
We have many very hard problems to solve before we can build anything that might
qualify as an AGI. Many problems have become easier as computer power has
increased on its exponential and seemingly inexorable way. But we also need
fundamental breakthroughs.
Consider four basic capabilities that any true AGI would have to possess:
The object-recognition capabilities of a 2-year-old child
The language capabilities of a 4-year-old child
The manual dexterity of a 6-year-old child
The social understanding of an 8-year-old child
What if there are some essential aspects of intelligence that we still do not
understand and that do not lend themselves to computation? We might need a new
conceptual framework. Creating a machine capable of effectively performing the
four capabilities above may take 10 years, or it may take 100.
My early work on robotic insects showed me the importance of coupling AI systems
to bodies. I don't see why, by the middle of this century, we shouldn't have
humanoid robots with agile legs and dexterous arms and hands.
I believe the AGIs of the future will not only be able to act intelligently but
also convey emotions, intentions, and free will. In fact, one of my dreams is to
develop a robot that people feel bad about switching off, as if they were
extinguishing a life.
Maybe some kind of AGI already exists on the Google servers, probably the single
biggest network of computers on our planet, and we aren't aware of it. At the
2007 Singularity Summit, I asked Peter Norvig, Google's chief scientist, if the
company had noticed any unexpected emergent properties in its network. He
replied that they had not seen anything like that.
Will machines become smarter than us and decide to take over? I don't think so.
To begin with, there will be no "us" for them to take over from. We are already
starting to change ourselves from purely biological entities into mixtures of
biology and technology. We are more likely to see a merger of ourselves and our
robots before we see a standalone superhuman intelligence.
While we become more robotic, our robots will become more biological, with parts
made of artificial and yet organic materials. In the future, we might share some
parts with our robots.
We need not fear our machines because we will always be a step ahead of them,
because we will adopt the new technologies used to build those machines right
into our own heads and bodies.
Rodney Brooks is a professor at the Massachusetts Institute of Technology. He
researches the engineering of intelligent robots capable of operating in
real-world environments and how to understand human intelligence by building
humanoid robots. Brooks is also the chief technical officer of iRobot Corp.
Ray Kurzweil and Neil Gershenfeld
MIT professor Neil Gershenfeld and technology futurist Ray
Kurzweil have long worked at the leading edges of physical science and computer
science. Both believe that we are on the event horizon of a technological
singularity.
David Dalrymple, now age 16 and an MIT graduate student, began corresponding
with Gershenfeld in 1999. Later that year, Gershenfeld invited him to a White
House event to demonstrate a device he had built using Lego Mindstorms. There
Dalrymple met Kurzweil. Dalrymple worked with Kurzweil for three summers as an
undergraduate. He graduated at age 13 and is now working toward his Ph.D. under
Gershenfeld.
Gershenfeld, director of MIT's Center for Bits and Atoms, studies the boundary
between computer science and physical science. Kurzweil has been fascinated with
modeling the physical world in computers, and believes he may just survive long
enough to see computers that are far smarter than people.
For years, Dalrymple has been trying to reconcile these two visions of the
future: Gershenfeld's future in which computers collapse and simply become part
of reality, and Kurzweil's future in which reality as we know it collapses and
simply becomes part of computers.
Ray Kurzweil
"We see these apparently opposing trends in many contexts.
Studying natural intelligence gives us the insights to create artificial
intelligence while at the same time artificial intelligence is extending our
natural intelligence. Reverse-engineering biology is giving us creative new
designs for advanced technologies, while those same technologies overcome the
limitations of biology."
"We will be infusing physical reality with embedded, distributed,
self-organizing computation everywhere. And at the same time we will be using
these massive and exponentially expanding computational resources to create
increasingly realistic, full-immersion virtual reality environments that compete
with and ultimately replace real reality."
Neil Gershenfeld
"I had always considered Ray and me to be headed in opposite
directions. He developed artificial intelligence and virtual worlds while I was
interested in the natural intelligence of physical systems. He forecast the
future while I was investigating technologies that are possible in the present."
"The result for me has been an increasingly close integration of physical
science and computer science, bringing the programmability of the digital world
to the physical world. But whether computers are merged with reality or reality
is merged with computers, the result is the same: the boundary between bits and
atoms disappears."
Signs of the Singularity
I think it's likely that with technology we can in the fairly
near future create or become creatures of more than human intelligence. Such a
technological singularity would revolutionize our world, ushering in a posthuman
epoch. In my 1993 essay, "The Coming Technological Singularity," I said I'd be
surprised if the singularity had not happened by 2030. I'll stand by that claim,
assuming we avoid showstopping catastrophes.
I expect the singularity will come as some combination of the following:
The AI Scenario: We create superhuman artificial intelligence (AI) in computers.
The IA Scenario: We enhance human intelligence through human-to-computer interfaces to achieve intelligence amplification (IA).
The Biomedical Scenario: We directly increase our intelligence by improving the neurological operation of our brains.
The Internet Scenario: Humanity, its networks, computers, and databases become sufficiently effective to be considered a superhuman being.
The Digital Gaia Scenario: The network of embedded microprocessors becomes sufficiently effective to be considered a superhuman being.
Philosopher Alfred Nordmann criticizes the extrapolations used to
argue for the singularity. Using trends for outright forecasting is asking for
embarrassment. And yet there are a couple of trends that at least raise the
possibility of the technological singularity. The first is a very long-term
trend, namely life's tendency toward greater complexity.
In the last few thousand years, humans have begun the next step, creating tools
to support cognitive function. We're building machines and systems that can
speed up the processes of problem solving and adaptation.
In recent decades, the enthusiasts have been encouraged by an enabling trend:
the exponential improvement in computer hardware. If the economic demand for
improved hardware continues, it looks like Moore's Law can continue for some
time. Moore's Law enables improvement in communications, embedded logic,
information storage, planning, and design. As long as the software people can
successfully exploit Moore's Law, the demand for this progress should continue.
Roboticist Hans Moravec may have been the first to draw a numerical connection
between computer hardware trends and artificial intelligence. Writing in 1988,
Moravec took his estimate of the raw computational power of the brain together
with the rate of improvement in computer power and projected that by 2010
computer hardware would be available to support roughly human levels of
performance.
Rodney Brooks suggests that computation may not even be the right metaphor for
what the brain does. If we are profoundly off the mark about the nature of
thought, then this objection could be a showstopper. But research that might
lead to the singularity covers a much broader range than formal computation.
Consider economist Robin Hanson's "shoreline" metaphor for the boundary between
those tasks that can be done by machines and those that can be done only by
human beings. Once upon a time, there was a continent of human-only tasks. By
the end of the 1900s, that continent had become an archipelago. We might recast
much of our discussion in terms of the question, "Is any place on the
archipelago safe from further inundation?"
The goal of enhancing human intelligence through human-computer interfaces (the
IA Scenario) is near. Today a well-trained person with a suitably provisioned
computer can look very smart indeed. Consider just a slightly more advanced
setup, in which an Internet search capability plus math and modeling systems are
integrated with a head-up display. The resulting overlays could give the user a
kind of synthetic intuition about his or her surroundings.
The Biomedical Scenario directly improving the functioning of our own brains
has a lot of similarities to the IA Scenario, though computers would be only
indirectly involved, in support of bioinformatics. In the near future, drugs for
athletic ability may be only a small problem compared with drugs for intellect.
Brooks suggests that the singularity might happen and yet we might not notice. I
think a pure Internet Scenario, where humanity plus its networks and databases
become a superhuman being, is the most likely to leave room to argue about
whether the singularity has happened or not. In this future, there might be no
explicit evidence of a superhuman player.
The Digital Gaia Scenario would probably be less deniable, if only because of
the palpable strangeness of the everyday world: reality itself would wake up.
The Digital Gaia would be something beyond human intelligence, but nothing like
human. Digital Gaia is a hint of how alien the possibilities are.
The best answer to the question, "Will computers ever be as smart as humans?" is
probably "Yes, but only briefly."
Consciousness, intelligence, self-awareness, emotion even their definitions
have been debated since forever. Now there is the possibility of making progress
with these mysteries. Some of the hardest questions may be ill-posed, but we
should see a continuing stream of partial answers and surprises. Each partial
success is removing more dross, closing in on the ineffable features of mind. Of
course, we may remove and remove and find that ultimately we are left with
nothing but devices that are smarter than we are and that is the singularity.
Vernor Vinge first used the term singularity to refer to the advent of
superhuman intelligence while on a panel at the annual conference of the
Association for the Advancement of Artificial Intelligence in 1982. Three of his
books won the Hugo Award for best science-fiction novel of the year. From 1972
to 2000, Vinge taught math and computer science at San Diego State University.
Tech Luminaries Address Singularity
Douglas Hofstadter
Pioneer in computer modeling of mental processes; director of the
Center for Research on Concepts and Cognition at Indiana University,
Bloomington; winner of the 1980 Pulitzer Prize for general nonfiction.
"It might happen someday, but I think life and intelligence are far more complex
than the current singularitarians seem to believe, so I doubt it will happen in
the next couple of centuries. [The ramifications] will be enormous, since the
highest form of sentient beings on the planet will no longer be human."
Jeff Hawkins
Cofounder of Numenta, in Menlo Park, California, a company
developing a computer memory system based on the human neocortex. Also founded
Palm Computing, Handspring, and the Redwood Center for Theoretical Neuroscience.
"If you define the singularity as a point in time when intelligent machines are
designing intelligent machines in such a way that machines get extremely
intelligent in a short period of time an exponential increase in intelligence
then it will never happen. Intelligence is largely defined by experience and
training, not just by brain size or algorithms."
"Machines will understand the world using the same methods humans do; they will
be creative. Some will be self-aware, they will communicate via language, and
humans will recognize that machines have these qualities."
"I don't like the term 'singularity' when applied to technology. A singularity
is a state where physical laws no longer apply because some value or metric goes
to infinity, such as the curvature of space-time at the center of a black hole.
No one can predict what happens at a singularity."
"Exponential growth requires the exponential consumption of resources (matter,
energy, and time), and there are always limits to this. Why should we think
intelligent machines would be different? We will build machines that are more
'intelligent' than humans, and this might happen quickly, but there will be no
singularity, no runaway growth in intelligence."
John Casti
Senior Research Scholar, the International Institute for Applied
Systems Analysis, in Laxenburg, Austria and cofounder of the Kenos Circle, a
Vienna-based society for exploration of the future. Builds computer simulations
of complex human systems. Author of popular books about science.
"I think it's scientifically and philosophically on sound footing. The only real
issue for me is the time frame over which the singularity will unfold. [The
singularity represents] the end of the supremacy of Homo sapiens as the dominant
species on planet Earth. At that point a new species appears, and humans and
machines will go their separate ways."
T.J. Rodgers
Founder and CEO of Cypress Semiconductor Corporation, in San
Jose, Californa. Owner of the Clos de la Tech winery and vineyards, in
California.
"I don't believe in technological singularities. It's like extraterrestrial life
if it were there, we would have seen it by now."
"I don't believe in the good old days. We live longer and better than our
predecessors did and that trend will continue in the future. We will also be
freer, more well educated and even smarter in the future but exponentially so,
not as a result of some singularity."
Eric Hahn
Serial entrepreneur and early-stage investor who founded Collabra
Software (sold to Netscape) and Lookout Software (sold to Microsoft) and backed
Red Hat, Loudcloud, and Zimbra. CTO of Netscape during the browser wars.
"I think that machine intelligence is one of the most exciting remaining 'great
problems' left in computer science. For all its promise, however, it pales
compared with the advances we could make in the next few decades in improving
the health and education of the existing human intelligences already on the
planet."
"I'm not worried about The Matrix or The Day the Earth Stood Still. But I do
hope the new intelligence doesn't run Windows."
Gordon Bell
Principal researcher at Microsoft Research, Silicon Valley. Led
the development of or helped design a long list of time-share computers and
minicomputers at Digital Equipment Corporation.
"Singularity is that point in time when computing is able to know all human and
natural-systems knowledge and exceed it in problem-solving capability with the
diminished need for humankind as we know it. I basically support the notion, but
I have trouble seeing the specific transitions or break points that let the
exponential take over and move to the next transition."
Steven Pinker
Professor of psychology at Harvard; previously taught in the
department of Brain and Cognitive Sciences at MIT, with much of his research
addressing language development. Writes best sellers about the way the brain
works.
"There is not the slightest reason to believe in a coming singularity. The fact
that you can visualize a future in your imagination is not evidence that it is
likely or even possible. Look at domed cities, jet-pack commuting, underwater
cities, mile-high buildings, and nuclear-powered automobiles all staples of
futuristic fantasies when I was a child that have never arrived. Sheer
processing power is not a pixie dust that magically solves all your problems."
Gordon E. Moore
Cofounder and chairman emeritus of Intel Corporation, cofounder
of Fairchild Semiconductor, winner of the 2008 IEEE Medal of Honor, chairman of
the board of the Gordon and Betty Moore Foundation. The Moore of Moore's Law.
"I am a skeptic. I don't believe this kind of thing is likely to happen, at
least for a long time. And I don't know why I feel that way. The development of
humans, what evolution has come up with, involves a lot more than just the
intellectual capability. You can manipulate your fingers and other parts of your
body. I don't see how machines are going to overcome that overall gap, to reach
that level of complexity, even if we get them so they're intellectually more
capable than humans."
Jim Fruchterman
Founder and CEO of the Benetech Initiative, in Palo Alto, one of
the first companies to focus on social entrepreneurship. Former rocket scientist
and optical-character-recognition pioneer. Winner of a 2006 MacArthur
Fellowship.
"I believe the singularity theory is plausible in that there will be a major
shift in the rate of technology change."
"I think that futurists are much more successful in projecting simple measures
of progress (such as Moore's Law) than they are in projecting changes in human
society and experience."
Esther Dyson
Commentator and evangelist for emerging technologies, investor
and board member for start-ups.
"The singularity I'm interested in will come from biology rather than machines."
AR Thanks, IEEE, for a great report. This was the theme in my 1996 novel Lifeball, where I called the singularity syzygy. As for all the consciousness stuff, see my 2004 book Mindworlds. And as for my first reactions to Kurzweil singularitarianism, see my 2006 robots article plus responses by Stephen Robbins and Michael Zimmerman.
