James Miller, an econ professor at Smith and the author of Singularity Rising, asked if he could present his case for
the future likelihood of a Singularity. Or, as the Original Cyberpunk has described it, “the rapture of the nerds”. Since this is a place where we are always pleased to give both space and genuine consideration to diverse points of view, I readily agreed to his request.
I define a Singularity as a threshold of time at which AIs at least as
smart as humans and/or augmented human intelligence radically remake
civilization.1. Rocks exist!
Strange as it seems, the existence of rocks actually
provides us with evidence that it should be possible to build computers
powerful enough to take us to a Singularity.
There are around ten trillion, trillion atoms in a one-kilogram rock,
and as inventor and leading Singularity scholar Ray Kurzweil writes: “Despite the apparent solidity of the object, the atoms are
all in motion, sharing electrons back and forth, changing particle spins, and
generating rapidly moving electromagnetic fields. All of this activity represents computation,
even if not very meaningfully organized.”If the particles in the rock were organized in a more
“purposeful manner” it would be possible to create a computer trillions of
times more computationally powerful than all the human brains on earth
combined. Our eventual capacity to
accomplish this is established by our second fact.2. Biological cells exist!
The human body makes use of tiny biological machines to
create and repair cells. Once mankind
masters this nanotechnology we will be able to cheaply create powerful
molecular computers. Our third fact
proves that these computers could be turned into general purpose thinking
machines.3. Human brains exist!
Suppose this book claimed that scientists would soon build a
human teleportation device. Given that
many past predictions of scientific miracles—such as cheap fusion power, flying
cars or a cure for cancer—have come up short, you would rightly be suspicious
of my teleportation prediction. But my
credibility would jump if I discovered a species of apes that had the inborn
ability to instantly transport themselves across great distances.In some alternate universe that had different laws of
physics, it’s perfectly possible that intelligent machines couldn’t be created. But human brains provide absolute proof that
our universe allows the construction of intelligent, self-aware machines. And, because the brain exists already,
scientists can probe, dissect, scan and interrogate it. We’re even beginning to understand the
brain’s DNA and protein-based ‘source code’.
Also, many of the tools used to study the brain have been becoming
exponentially more powerful, which explains why engineers might be only a
couple of decades away from building a working digital model of the brain even
though today we seem far from understanding all of the brains operations. Would-be creators of AI are already using
neuroscience research to help them create machine learning software. Our fourth fact shows the fantastic
potential of AI.4. Albert Einstein existed!
It’s extremely unlikely that the chaotic forces of evolution
just happened to stumble on the best possible recipe for intelligence when they
created our brains, especially since our brains have many constraints imposed
on them by biology: they must run on energy obtained from mere food; must fit
in a small space; and can’t use useful materials such as metals and plastics,
that engineers employ all the time.But even if people such as Albert Einstein had close to the
highest possible level of intelligence allowed by the laws of physics, creating
a few million people or machines possessing this man’s brain power would still
change the world far more than the industrial revolution. We share about 98% of our genes with some primates, but that
2% difference was enough to produce creatures that can assemble spaceships,
sequence genes, and build hydrogen bombs.
What happens when mankind takes its next step, and births lifeforms who
have a 2% genetic distance from us?5. If we were smarter, we would be smarter!
Becoming smarter enhances our ability to do everything,
including our ability to figure out ways of becoming even smarter because our
intelligence is a reflective superpower able to turn on itself to decipher its
own workings. Consider, for example, a
college student taking a focus-improving drug such as Adderall, Ritalin or
Provigil, to help learn genetics. After
graduation, this student might get a job researching the genetic basis of human
intelligence, and her work might assist pharmaceutical companies in making
better cognitive enhancing drugs that will help future students acquire an even
deeper understanding of genetics.
Smarter scientists could invent ways of making even smarter scientists
who could in turn… Now, throw the power of machine intelligence into this
positive feedback loop and we will end up at technological heights beyond our
imagination.
I hereby recuse myself from the position of critic, mostly since my position on the concept can be best described as “mild, but curious skepticism”. But everyone should feel free to either express their doubts or offer additional arguments to bolster Prof. Miller’s case.