TIME: Almanac 1993/TIME Almanac 1993.iso/time/062590/0625630.000

Text File | 1992-08-28 | 6.1 KB | 130 lines

                                                                                IDEAS, Page 74

Those Computers Are Dummies



A physicist's attack riles artificial-intelligence researchers

By MICHAEL D. LEMONICK


    Roger Penrose is hardly the sort of man who would normally
excite much popular interest, let alone controversy. The shy,
somewhat rumpled and unfailingly polite Oxford professor, 58,
has spent most of his career spinning theories in the most
abstruse areas of mathematics and physics. His contributions
to both disciplines have earned him a sterling reputation among
his colleagues. But his pursuits have been so far removed from
the everyday world that few people outside his fields of
expertise were even aware of his existence.

    Until last fall, that is, when Penrose brought out the book
The Emperor's New Mind, an in-depth discussion of the
relationship between artificial intelligence, consciousness and
the laws of physics. Despite its complexity and intellectual
rigor, the book quickly jumped onto the best-seller list. And
just like his friend and sometime collaborator Stephen Hawking,
the once obscure Penrose suddenly found himself showered with
publicity. The professor was so unaware of how much the book
was earning that he asked his editor whether there was enough
to cover a few thousand pounds for a new car.

    In the ensuing months, he has also found himself under
attack. Reason: Penrose's central conclusion is that computers
will never think because the laws of nature do not allow it.
That angers many artificial-intelligence researchers. M.I.T.'s
Marvin Minsky, one of the field's pioneers, is downright
hostile. Says he: "Penrose is O.K. when he talks about
mathematics, but most of his evidence argues against his
conclusions. As far as I can tell, he is just plain wrong."
Stanford psychologist and AI researcher David Rumelhart is
somewhat milder: "He defines intelligence too narrowly by
saying it depends on consciousness."

    Nonetheless, Penrose's reasoning is powerful, and he delves
extensively into such heavy topics as fractal geometry, number
theory, quantum physics, entropy and cosmology to give readers
the necessary background to understand his ideas. "I have to
admit," he says, "that I had been looking for an excuse to
write about many aspects of physics and mathematics anyway, and
this gave me one."

    Penrose's first major point is that the human mind can reach
insights that are forever inaccessible to computers. The reason
is that all digital computers operate according to algorithms,
or sets of rules that prescribe how to solve problems. Yet
there are problems that cannot be approached by any system of
rules, a fact shown in the 1930s by the mathematician Kurt
Godel. Godel's theorem establishes that in any mathematical
system there must be certain propositions that are obviously
true but that can never be proved within the rules of the
system.

    Mathematician Alan Turing made a related discovery in the
1950s when he used his Turing machine -- an imaginary, simple
computer -- to prove that there are some mathematical problems
that are solvable but that cannot be solved even in principle
by a digital computer. Says Penrose: "The very fact that the
mind leads us to truths that are not computable convinces me
that a computer can never duplicate the mind" -- this despite
the fact that the human brain is often described as a
particularly complex computer.

    As Penrose freely admits, his other main theory is far more
speculative. It holds that consciousness and insight, which he
says are beyond the capabilities of computers, are governed by
as yet undiscovered laws of physics. The idea that computers
are necessarily unconscious and without insight is largely
based on his own experience in solving abstract puzzles. And
it is true that these mental processes are not explained by
existing laws of physics. The answers will come, says Penrose,
with the merger of Einstein's theory of relativity, which
concerns itself with gravity, and quantum theory, which governs
the sub-microscopic world. These two theories are
mathematically incompatible, and physicists are hard at work
trying to create a quantum version of gravity.

    One consequence could be to establish the boundaries of
quantum mechanics, which says particles can suddenly jump from
one place to another without traversing the space in between.
Penrose's intuition, although he has no proof, is that these
effects may apply not just to atoms but also to objects as big
as brain cells. An act of creative thinking, he argues, could
be the outward manifestation of neurons making quantum jumps
from one energy state to another. Since computers do not operate
by quantum rules, he says, they will never have insights.

    Moreover, he believes, quantum gravity could be behind
consciousness itself. The argument is tricky, but one reason
for this belief is that consciousness carries with it a
peculiarity that baffles physics: humans perceive time as
moving forward rather than backward. But virtually all the laws
of physics are time symmetric -- they work equally well in
forward or in reverse -- and the mind presumably operates by
physical laws.

    Penrose's answer is that when quantum gravity is finally
constructed, it will prove to be time asymmetric -- that is,
it will not work in reverse. Why? Because the Big Bang that
started the universe must, at its earliest moments, have been
governed by quantum gravity. And the Big Bang was surely a
time-asymmetric phenomenon that could not happen in reverse.
If quantum gravity winds up being the theory governing the
mind, that will also explain why time moves forward, not
backward.

    In short, Penrose believes human creativity and
consciousness are nothing less than the perceptible workings
of the most basic laws of the universe. It is a bolder position
than other physicists are prepared to take, but Penrose likes
to be different. Says he: "Worrying about things that no one
else worries about is where insights come from."