IBM scientist and Watson project director Dave Ferrucci stands with Watson, which will compete against human contestants this fall on "Jeopardy!"
Despite what a good etymologist might tell you, our society doesn’t find trivia particularly trivial. Our brainiest high school kids compete in Quiz Bowl Tournaments. Our rowdiest bar patrons cry foul when the trivia night host gives the other team easy questions. And our oldest technology corporation has chosen Jeopardy! as the gold standard by which to judge its A.I.
I.B.M.’s new supercomputer, Watson, is a sibling of sorts to the chess-playing Deep Blue. The New York Times Magazine has the story of Watson’s development, and says it’s set to compete in Jeopardy! sometime next fall.
Watson’s potential to dominate a trivia competition got me thinking about Ken Jennings. In college, my friends and I watched as Jennings won 74 straight Jeopardy! games, often by wide margins. It was just six years ago, but trivia was different back in Jennings’ day.
Back then, trivia wasn’t confined to game shows or organized competitions. Sitting in a park or at a bar, we might have tried to remember some trivial information, like where the guitar originated or how Babe Ruth died. If nobody knew, we’d argue, make guesses, and finally settle on a likely answer.
This kind of informal trivia still happens, but I’ve notice it usually ends with someone pulling out a smart phone and getting the answer with little drama or discussion.
Unlike us and our smart phones, Watson is not connected to the Internet. It depends on the millions of pages of documents stored in its databanks —novels, encyclopedias, religious texts — to answer questions. As the New York Times pointed out, it abides by the same rules as Jeopardy! contestants, only using what it has in its memory.
But Watson is not an A.I marvel because of its large databanks. Watson is impressive because it can decode the wordplay and subtlety of a Jeopardy! question, create a list of probable answers, and then and produce one specific response it deems most likely correct.
Watson’s ranking of likely answers reminds me of how an informal trivia question used to be handled. And something at work the other day illustrates how things are different. We were talking about the president who came before Lincoln. Apparently Congress hated him so much that it refused to pay for his official White House portrait.
But who was the president before Lincoln, anyway? No one could remember, so we turned back to our computers. I typed the question into Google and had the answer in seconds: James Buchanan.
But what I didn’t have was the fun of hashing out the possible answers and deciding on one, even if it might not be correct.