Watson vs. Mankind: Round 1

Ben Templeton's take on Round 1 of the knock down, drag out fight between artificially and genuinely intelligent beings.


- 44 Maagnum


Night 1 of Watson vs. Ken Jennings vs. Brad Rutter (anybody else think Brad looks a bit like Hans Gruber from Die Hard?) was exciting and entertaining. Watson started off well, jumping out to a early lead of $5200 while the next closest was $1000. As time went on, the humans pulled a bit back, as Watson gave some amusing errors. The most laughs came when Mr. Jennings gave “the 20’s” as the incorrect response to a clue, Watson buzzed in and suggested “What is the 1920’s?” The night ended with Watson and Mr. Rutter tied at $5000, and Ken trailing distantly with $2000.

I’m going to do a little bit on each night, providing a little commentary, “Carpe Daemon”-style. Maybe some people will tune in tomorrow night who missed tonight, after seeing the frenzy of media coverage (does this blog post technically make me a part of the media?). And they certainly should, if you ask me. It’s difficult to overstate the significance of this achievement, for society and technology. (If you want a quick rundown of the importance of this shindig, check out this post from a few days ago.)

Brad Rutter

I’m going to talk a little bit about what was probably the moment that most people will remember, which was Watson’s repetition of Ken’s incorrect response. The “answer” came from a category about “the decade in which X, Y, and Z occurred”, and Ken’s question was “what is the 20’s?” As the TV audience, we got to see Watson’s confidence values displayed on the screen. Without confidence above a certain threshold, Watson will keep doing calculations until it gets high enough. And in this case, we saw that Watson was certain that the answer was the 1920’s. There was a moment of pause as we waited to see if it was going for the buzzer. And then it did, making a fool of itself to anybody who managed to anthropomorphize Watson sufficiently (which one author certainly did, writing an article that consisted entirely of criticism of the aesthetic presentation of the supercomputer).

Hans Gruber

This moment was interesting and important for two reasons. First, I think it epitomized the difficulties of artificial intelligence. Watson’s capacity to err in a situation that is so trivial to human intelligence shows us how difficult the littlest things can be. In this case, perhaps the engineers could included a microphone and some voice recognition technology. But the problem of turning sound into words and the symbolic logic that represents them is far from trivial. The point is that to a computer, nothing comes easy, even things so simple as to make the audience chuckle a bit.

The second important idea to take from that moment is that while Watson got the wrong answer, it got the same wrong answer as Ken Jennings, who won Jeopardy 74 times in a row. I personally think that it’s very cool and pretty impressive that through 4 years of hard work and piles of electronic logic, the engineers and programmers at IBM managed to produce a machine that thinks in the same way as one of the human minds that is most adept at synthesizing information into knowledge, that of Mr. Jennings. And that accomplishment is far from trivial.

Tune in tomorrow at 7:00 for the next installment in this historic saga, on ABC. Seriously.