Watson, Jeopardy, and intelligent machines

June 16th, 2010

Software firms and university scientists have produced question-answering systems for years, but these have mostly been limited to simply phrased questions. Nobody ever tackled “Jeopardy!” because experts assumed that even for the latest artificial intelligence, the game was simply too hard: the clues are too puzzling and allusive, and the breadth of trivia is too wide.

With Watson, I.B.M. claims it has cracked the problem — and aims to prove as much on national TV. The producers of “Jeopardy!” have agreed to pit Watson against some of the game’s best former players as early as this fall.

The New York Times profiles IBM’s Watson. It’s both a look at how far AI systems have come, and how far they still have to go. I’d be really interested in reading more about the underlying algorithms, so that I can get a better idea of where the major bottlenecks are.

It’s going to be amazing when we can apply machines like this to tasks like decision making in hospitals, or even create new, more intuitive, data mining interfaces. Hell, if we start cross-breeding Watson with something like ADAM, the robot that can form and test hypotheses, us graduate students may be made obsolete.

Well, someday…

Tags: , | 2 Comments »

2 Responses to “Watson, Jeopardy, and intelligent machines”

  1. I won three games in 1990 – not by any stretch of the imagination a super player. I’d venture to guess the machine could beat me just on reaction time alone.

  2. Oddly enough, reaction time seems to be one of the machine’s weak points. As I’m sure you’re aware, answering a question incorrectly is the worst thing you can possibly do in Jeopardy. Thus, they’ve programmed it not to ring in until it’s reasonably sure that it has a correct answer. This puts it at a disadvantage, because often humans will ring in on gut instinct, then use the 5 seconds of time to actually come up with their response.