DAVID GREENE, HOST:
There was an epic battle that took place last night at the IBM offices in San Francisco. A computer formally debated a person in what some experts are billing as a breakthrough in artificial intelligence. NPR's Laura Sydell was there, and she's with us. Hi, Laura.
LAURA SYDELL, BYLINE: Hello. Good morning.
GREENE: So this sounds like something out of a sci-fi movie. Was it as cool as it sounds?
SYDELL: Well, you know, it was just this person standing next to this obelisk-shaped computer thing so it didn't look very interesting.
GREENE: (Laughter) OK.
SYDELL: You know, truthfully.
GREENE: Well, then why did IBM want to do this? What was this about?
SYDELL: Well, it was to show off some actually pretty cool new computer skills. You know, IBM has made news in the past for, you know, a computer that beat a world champion in chess, and it also beat champions at the game "Jeopardy!" This is a little different because in a debate, there aren't clear rules. There isn't a numerical score. So they built this computer. It's called Project Debater, and they put it up against two experienced debaters.
GREENE: OK. So the computer was actually talking to two people? I mean going back and forth debating something.
SYDELL: That is right.
SYDELL: It actually spoke, you know, for four minutes and then for two minutes. It went through a formal debating process.
GREENE: Can we say who won? Did the computer beat the humans?
SYDELL: Well, you know, actually I'd say it was a draw in the sense that there were two debates and they pulled the audience, which was largely journalists, ahead of time to see how they felt about a particular topic, and then they pulled them after. And in one case, the computer actually did in fact beat the human. And the computer argued that telemedicine - that was the topic of the debate - was a good thing. And there was kind of a funny moment where the human debaters said that they didn't think that telemedicine was good - this is when doctors treat people from a distance - because you didn't have the physical hand of the doctor or the nurse, and the computer responded with some humor. It said, I am a true believer in the power of technology, as I should be.
GREENE: As I should be. (Laughter) Wow. The computer making a joke, that's amazing.
SYDELL: It's pretty good. I did speak with one of the debaters, Noa Ovadia, and she was relatively impressed with the computer.
NOA OVADIA: I think the machine has reached an incredible degree of fluency, both in terms of sentence construction and in terms of argument construction. And I think it's at the level where it's comparable to some average debaters. Maybe not the best in the world, but definitely well-informed reasoning and logical construction.
GREENE: So is this a big moment? I mean, is artificial intelligence, has it totally arrived? Are computers definitely getting as smart as us, or smarter?
SYDELL: Well, I think it's significant in showing that a computer can create a coherent argument and pull together information and make a case. I mean, the computer didn't know what the debate would be ahead of time. OK? It had to actually sort through millions of documents and look for language that would help it make its case in real time. So mostly the language it used came directly from text. And ultimately what this means is that we could start to use computers - say, a lawyer could use it to make better arguments, or it might help doctors and patients decide on the best course of treatment or it could help you even pick stocks.
GREENE: Wow. Can I get one of these computers anytime soon?
SYDELL: Well, alas, it's not going to come out into the consumer market. You won't be fighting yet with Alexa and Siri (laughter).
GREENE: (Laughter) OK. NPR's Laura Sydell on a computer debating a human. Laura, thanks.
SYDELL: You're welcome. Transcript provided by NPR, Copyright NPR.