Theos-Talk Archives (November 2003 Message tt00110)

Can Robots Become Conscious?

November 11, 2003
By KENNETH CHANG


It's a three-part question. What is consciousness? Can you
put it in a machine? And if you did, how could you ever
know for sure?

Unlike any other scientific topics, consciousness - the
first-person awareness of the world around - is truly in
the eye of the beholder. I know I am conscious. But how do
I know that you are?

Could it be that my colleagues, my friends, my editors, my
wife, my child, all the people I see on the streets of New
York are actually just mindless automatons who merely act
as if they were conscious human beings?

That would make this question moot.

Through logical analogy - I am a conscious human being, and therefore you as 
a human being are also likely to be conscious - I conclude I am probably not 
the only conscious being in a world of biological puppets. Extend the question 
of consciousness to other creatures, and uncertainty grows. Is a dog 
conscious? A turtle? A fly? An elm? A rock?

"We don't have the mythical consciousness meter," said Dr.
David J. Chalmers, a professor of philosophy and director
of the Center for Consciousness Studies at the University
of Arizona. "All we have directly to go on is behavior."

So without even a rudimentary understanding of what consciousness is, the 
idea of instilling it into a machine - or understanding how a machine might 
evolve consciousness - becomes almost unfathomable.

The field of artificial intelligence started out with dreams of making 
thinking - and possibly conscious - machines, but to date, its achievements have 
been modest. No one has yet produced a computer program that can pass the Turing 
test.

In 1950, Alan Turing, a pioneer in computer science, imagined that a computer 
could be considered intelligent when its responses were indistinguishable 
from those of a person. The field has evolved to focus more on solving practical 
problems like complex scheduling tasks than on
emulating human behavior.

But with the continuing gains in computing power, many believe that the 
original goals of artificial intelligence will be attainable within a few decades.

Some people, like Dr. Hans Moravec, a professor of robotics at Carnegie 
Mellon University in Pittsburgh, believe a human being is nothing more than a fancy 
machine, and that as technology advances, it will be possible to build a 
machine with the same features, that there is nothing magical about the brain and 
biological flesh.

"I'm confident we can build robots with behavior that is just as rich as 
human being behavior," he said. "You could quiz it as much as you like about its 
internal mental life, and it would answer as any human being."

To Dr. Moravec, if it acts conscious, it is. To ask more is pointless.

Dr. Chalmers regards consciousness as an ineffable trait, and it may be 
useless to try to pin it down. "We've got to admit something here is irreducible," 
he said. "Some primitive precursor consciousness could go all the way down" to 
the smallest, most primitive organisms, even
bacteria, he said.

Dr. Chalmers too sees nothing fundamentally different between a creature of 
flesh and blood and one of metal, plastics and electronic circuits. "I'm quite 
open to the idea that machines might eventually become conscious," he said, 
adding that it would be "equally weird."

And if a person gets into involved conversations with a robot about 
everything from Kant to baseball, "we'll be as practically certain they are conscious 
as other people," Dr. Chalmers said.

"Of course, that doesn't resolve the theoretical question," he said.

But others say machines, regardless of how complex, will never match people.

The arguments can become arcane. In his book "Shadows of the Mind," Dr. Roger 
Penrose, a mathematician at Oxford University in England, enlisted the 
incompleteness theorem in mathematics. He uses the theorem, which states that any 
system of theorems will invariably include statements that cannot be proven, to 
argue that any machine that uses computation - and hence all robots - will 
invariably fall short of the accomplishments of human mathematicians.

Instead, he argues that consciousness is an effect of quantum mechanics in 
tiny structures in the brain that exceeds the abilities of any computer.

http://www.nytimes.com/2003/11/11/science/11MACH.html?ex=1069533728&ei=1&en=b2
c8d5710ed32897

Copyright 2003 The New York Times Company 


[Non-text portions of this message have been removed]