My Problem with AI
It’s obvious to anybody that the mind does much more than solve problems.
David Gelernter is professor of computer science at Yale, chief scientist at Mirror Worlds Technologies, contributing editor at the Weekly Standard, and member of the National Council of the Arts. He is the author of several books and many technical articles, as well as essays, art criticism, and fiction. The "tuple spaces" introduced in Carriero and Gelernter's Linda system (1983) are the basis of many computer-communication and distributed programming systems worldwide. According to Reuters, his book "Mirror Worlds" (Oxford University Press, 1991) "foresaw" the World Wide Web and was "one of the inspirations for Java"; the "lifestreams" system (first implemented by Eric Freeman at Yale) is the basis for Mirror Worlds Technologies' software. Gelernter is also the author of "The Muse in the Machine" (Free Press, 1994), the novel "1939" (Harper Perennial, 1995), "Machine Beauty" (Basic Books, 1998), and most recently, "Judaism: A Way of Being" (Yale University Press, 2010).
My argument with the direction that AI has taken is that borrowing from the standard approaches in cognitive psychology and experimental psychology, AI had tended to say, “we want to know about thinking, then we’ll move onto emotion, or maybe consciousness.”
Thinking is what we want to do. The really important thing the mind does, the important activity the mind does is think, solve problems. A lot of people in AI used problem-solving as an equivalent of thinking.
They said, “Here’s what we’re working on. We’re working on artificial thought, artificial intelligence, problem-solving software.” But it’s obvious to anybody that the mind does much more than solve problems. It’s very rare for anybody to go about solving a problem formally.
Certainly we don’t’ do a lot of mathematical problem-solving, or problem sets in physicists most of the time. And to an extent that we are confronted with problems to solve, we almost always first have recourse to experience and we think, “well, what did I do the last time?”
In a more fundamental way, it is obvious to anybody that if I am working at my computer and I get tired and I lean back and look out the window and just watch the passing scene, I’m still thinking, my mind hasn’t shut down. I watch what’s happening. I react in more subtle cognitive ways to what I see. It’s obvious that when I get tired, when my mind starts to drift, when I move into the free associative state that was studied by Freud that we know precedes falling asleep - free association is a kind of thinking also.
My mind doesn’t shut off, but I’m certainly not solving problems. I’m wandering around. And we also know that when we sleep, we think also. Sleep thought is different than waking thought. Sleep thought is not solving problems in mathematics, or solving any kind of problems in a methodical way. Sleep thought is image thought, for the most part, and sleep thought is hallucinatory. I see things that aren’t there.
So we need to understand the connection, the spectrum that connects wide awake, focused, alert problem-solving type of thought with what happens to my mind as I get tired, as my focus decreases, as I approach sleep. Actually, the brain goes through several oscillations like this during the day, but there’s a continuous spectrum connecting my most focused, my sharpest kind of analytical thought, the sharpest of which I am capable of on the one hand, and the lowest focused kind of thought in which my mind drifts and ultimately I find myself asleep and dreaming.
The field of Artificial Intelligence had studied only the very top end of the spectrum and still tends to study only the very top end. It tends to say, what is thinking, it’s this highly focused, wide awake, alert, problem-solving state of mind. But not only is that not the whole story, but the problem, the biggest unsolved problem that has tended to haunt philosophy of mind, cognitive psychology, AI, is creativity.
People have always been fascinated about what makes for a creative person? What explains a creative leap, which is a well defined psychological event? People know when it happens to them. There is general agreement that to be creative is to have the ability to invent new analogies, to connect two things that are not obviously related. But once you have made the connection, you can see there is a relationship and other people can see the relationship too and creativity flow from that.
In Their Own Words is recorded in Big Think's studio.
Image courtesy of Shutterstock
Giving our solar system a "slap in the face."
- A stream of galactic debris is hurtling at us, pulling dark matter along with it
- It's traveling so quickly it's been described as a hurricane of dark matter
- Scientists are excited to set their particle detectors at the onslffaught
The climate change we're witnessing is more dramatic than we might think.
Once again, our circadian rhythm points the way.
- Seven individuals were locked inside a windowless, internetless room for 37 days.
- While at rest, they burned 130 more calories at 5 p.m. than at 5 a.m.
- Morning time again shown not to be the best time to eat.
SMARTER FASTER trademarks owned by The Big Think, Inc. All rights reserved.