When, Oh, When Will Computers Think?

How will we know beyond a doubt that a computer can call its own shots without being prompted by a human?

The question is not as simple as it sounds, and it likely won’t be answered in any of the ways we’re familiar with.

A statue of Alan Turing, the British mathematician who developed a method for identifying a thinking computer. The statue is made of stacked slate. Credit: zoonabar via Compfight cc

The Turing Test, a method of identifying thinking computers developed by British mathematician Alan Turing in 1950, may be the most famous technique.

It involves an interrogator in one room and a machine and a person in another. The object is for the interrogator to identify which is the person and which is the machine by gauging how they respond to questions.

The problem with the test, however, is that it assumes computer programmers will never get better at faking it.

Theoretical physicist Lawrence Krauss, as director of the Origins Project at Arizona State University, is exploring the most fundamental questions surrounding human existence, including consciousness.

Krauss said it’s not what a computer says but what it does that defines its consciousness.

“When a computer solves a problem, a problem you hadn’t asked it to solve, that’s when it will be sentient,” he said. “When it decides what problem is more interesting to solve.”

Sentience in this context refers to self-awareness, some degree of free will and a level of intelligence reflective of humans.

“This makes me think back to when a colleague of mine asked, ‘How will [sentient] computers do physics?’ They could perceive the world entirely differently,” he said. “Will they find the same problems interesting that we find interesting?”

That ability is the hallmark of an active mind, not a program, Krauss said.

The whole idea of sentience from programming is not at all startling to him. Krauss says in his book “A Universe From Nothing” that something can only come from nothing. It’s a good read if you’re reading this article.

Maybe he’s wrong, though. Maybe, as Fred Roberts, a senior engineer with Artificial Solutions said, “The unromantic answer is that [computers] are never going to think. Maybe they are just binary systems.”

Artificial Solutions sells software designed to facilitate natural-language communication between computers and people.

“We’re not trying to build sentience. We want to build a pleasant user experience,” said Roberts.

Then there’s Udo Middelmann, a scholar-in-residence at Nyack College and president of the Francis A. Schaeffer Foundation, a Christian think tank. He said a computer may reveal its sentience in a subtle way.

Middelmann gave the example of telling a computer to digitize a song from a vinyl record. A computer that decides on its own to also remix the audio — by effectively intuiting what the human wants — would suggest sentience, he said.

“When a computer fixes that for you, when it fulfills what you want based on its own reasoning without even being given the order, that would be a good indication,” he said.

Jim Nash is an award-winning business, tech and science journalist whose work has appeared in The New York Times, The Economist Group and Scientific American.

Jim Nash

Jim Nash

Contributor at Tech Page One
Jim Nash is an award-winning business, tech and science journalist whose work has appeared in The New York Times, The Economist Group and Scientific American.
Jim Nash
Tags: Technology