What’s up, Watson?

It’s been language, not math, that has our supercomputers stumped. Until now.

Julian Switala

Here’s some trivia for you: This term for a long-handled gardening tool can also mean an immoral pleasure seeker.

If you answered “What is a hoe?” then you’re in good company. Ken Jennings, the famous game show contestant who holds the record for the longest winning streak on “Jeopardy!” — 74 wins — answered “What is a hoe?” but his answer was rejected. The correct answer? “What is a rake?”

So you think you’re good at “Jeopardy!”? Meet Watson, the new IBM supercomputer. An expansive set of immobile servers may not seem intimidating, but appearances can be unreliable, deceptive and — in this case — downright frightening. Much like Deep Blue, a computer developed by IBM that defeated former world chess champion Gary Kasparov in a six-game chess match, Watson is also adept at an intellectually rigorous competition: “Jeopardy!”

For “Jeopardy!” players, the game is not particularly difficult. Players hear a question then provide an answer in question form. This process does not require any preparation nor does it entail much mental strain. Typically, players already have a bank of general knowledge with which they answer questions. In this regard, studying for “Jeopardy!” is pointless.

For humans, this process is incredibly straightforward and intuitive. Generally, we do not even think about a question or what it means; we just answer it. Of course, Watson is full of encyclopedic entries, almanac information and more data about the world than a group of humans could possibly memorize.

Yet Watson is still at a significant disadvantage. Why? Because programming a computer to understand natural language, or everyday colloquial speech, has been notoriously difficult for computer programmers and robotic linguists. This is especially true for a game like “Jeopardy!” which uses questions employing puns, double entendres, analogies, metaphors and clever conceptual relationships to the category under which questions fall.

To combat this difficulty, Watsonutilizes hundreds of linguistic algorithms in order to understand and attempt to correctly interpret “Jeopardy!” questions.

So why is IBM spending millions of dollars on such a machine? At first glance, Watson seems like a meaningless exercise in computer engineering. After all, it is being worked on with the ultimate objective of defeating humans on a televised episode of “Jeopardy!” which is really just an insignificant trivia game show.

However, the technology that Watson uses can be applied toward endeavors far more important than a mere game of “Jeopardy!”

Such a system would obviously transform the way search engines work by removing the necessity of searching for key terms. Even more revolutionary is that a perfected system would nullify the necessity of creating individualized work. Imagine when you will be able to type any of your college essay questions to Watson and then receive several options from which to choose. Or, if you just so happened to be a columnist for The Minnesota Daily, you could ask Watson to write for you. Despite the fact that I would like to keep my job, this prospect is awesome, frighteningly and terrifyingly awesome.

Julian Switala welcomes comments at [email protected]