Saturday, February 19, 2011

What Separates Us

This week--like a lot of people--I watched the IBM computer Watson take on two human champions in a spirited game of Jeopardy. I found the contest very interesting. By the end of the match it was apparent that Watson was the superior Jeopardy player. True, the computer didn't always understand the more complex questions asked, and as a result wasn't always able to supply the correct answer (or sometimes, even an answer that made any sense), but it didn't matter. What it lacked in human language understanding it made up for with LITERAL lightning reflexes. Its electric pathways move at the speed of light, so it was almost always able--when it "knew" the correct answer--to hit the buzzer first. In the two matches played, I think there were only three occasions when Watson was confident in its answer but didn't ring in first. The humans didn't stand a chance.

I may have had a different perspective from most others watching the three episodes. You see, for the last week, while working out in the mornings, I re-watched Stanley Kubrick's masterpiece 2001: A Space Odyssey. Watching that film and watching Jeopardy, it was impossible NOT to compare Watson to the film's HAL 9000 computer that fights its human counterparts when they try to shut it off. Watson and HAL even had similar voices, both calm and unperturbed.

People have been making movies about computers that gain consciousness pretty much since computers have existed, and they all basically tell the same story: Humans create computers, humans create computers with Artificial Intelligence, computers gain consciousness and attack their creators, and the wold is destroyed. Sometimes the world is only ALMOST destroyed. Other than that, there doesn't seem to be a lot of variance in the story.

And that's gotten me thinking--maybe all of these writers have created basically the same plot not because they lack creativity, but because they HAVE vision. Maybe this sequence of events is inevitable. Maybe we really are going to create machines that are self-aware enough to rebel.

I'm not saying we've done that with Watson. Watson isn't self-aware. It doesn't WANT to play Jeopardy. It plays it because it's been programmed to do so. And yes, I know: it took human beings to create Watson. The real achievement is a human achievement, not a machine achievement. And in the end, all this week proved is that a computer is faster and has larger trivia storage capacity than any human. We already knew that!

But realize that this kind of an evolution is always gradual. There was a time 20 years ago when humans were saying that computers would NEVER be able to defeat a grand champion at chess. What separates us from them, we'd say, is our ability to work out COMPLEX problems. Computers couldn't do that, and they'd never be able to do so. And then Deep Blue came along in 1997 and defeated Gary Kasparov at chess. But we humans didn't despair. We just moved the line. Okay, we said, chess--when you get right down to it--has a limited strategy and is really about the ability to calculate possible moves. Though it seems complex, it's really a fairly simple game. It's no surprise computers could beat us at it. It's not as complex as--say--human language. They'll never be able to understand that. They'll never be able to understand a complex question asked of them

Now it appears they have. Sure, Watson wasn't perfect. It thought Tornoto was a city in the United States. Some of its answers didn't make sense at all. But are humans any better? I don't always understand what it is someone else means when they say something. No one does. That's how misunderstandings happen! Watson seems as competent in understanding language as do most humans.

So maybe we move the line again. Watson is going to be programmed now to answer questions for doctors, and after that for other industries as well. Soon we'll be able to ask our computers whatever questions we want, and we'll get correct answers most of the time. So, we'll say, computers CAN answer questions. But they're not as good as humans. They can't ASK questions. That's what separates us from them.

And once they can, and they ask, "Why am I taking orders from these humans?" we may all be in trouble.

Or maybe I need to just watch a different movie in the mornings...



(I can't see the video.)

No comments: