Distracted by the buzzer

It’s been an interesting week for students of intelligence.  A supercomputer known as Watson competed in an exhibition Jeopardy match against two human players, Ken Jennings and Brad Rutter.  The human players were a couple of the best to ever play the game, but they were defeated.  Conceding the success of the computer in his Final Jeopardy question, Jennings, wrote “I, for one, welcome our new computer overlords.” 

The response is the normal one that arises when a machine appears to best humans at their own game: the computer cheated through an unfair advantage.  Many commentators have noted that the machine did not have to listen to the questions, but instead received them electronically and had an unfair advantage by being able to buzz in first.   Of course at the top level Jeopardy has always been won by whoever could buzz in first; the players generally know the answers, and the winner is the one who can get Trebek’s attention first.  Mr. Jennings and Mr. Rutter are renowned for this quick signaling ability – Jennings appeared on Jeopardy 74 straight times, in large part because he could signal his answers faster than any of his opponents.   So the electromechanical signaling interface of Watson is quicker than the humans.  No surprise, and not very interesting.

What is interesting is that Watson can take the strange “questions” of Jeopardy, generate potential answers, rank them, and quickly offer a best choice.   Even when Watson was not first on the buzzer, it usually had a correct answer to offer.  The capability to dissect a sentence, find important words, analyze the grammar, and connect that with other information (for example,  the name of the category in Jeopardy) is a significant one.  That Watson was able to do it quickly and quite well suggests many possible advances and applications.  These range from the trivial -- a phone menu that works -- to the significant -- rapid, automated, feedback for students in a personal learning environment, or improved free text search.