Unless movies in which robots with advanced intelligence pose a threat to humankind give you nightmares, you may very well be delighted, or at least intrigued, by IBM's "Jeopardy"-playing supercomputer Watson.
Watson has already proven it can defeat human "Jeopardy" champs in head-to-head combat. In an exhibition round today with 15 questions, the machine racked up $4,400 while organic life form competitor Ken Jennings earned $3,400 and Brad Rutter scored $1,200. Time will tell whether Watson can fare as well during a full game of "Jeopardy" in February.
In the meantime, it's worthwhile to consider the implications of a computer with sufficient artificial intelligence to ably play a game like "Jeopardy" that not only tests the amount of raw data a machine (or human) has stored away, but also its ability to analyze natural language -- "Jeopardy" categories and answers contain puns, for example -- so as to understand what sort of information is really being requested and to present that information clearly, concisely, and quickly.
The effects of a "Jeopardy"-playing supercomputer for the business or academic world is not lost on IBM, of course. As IBM Research Program Manager David Shepler put it, "IBM is not in the entertainment business. But we are in the business of technology and pushing frontiers."
In other words, "Jeopardy" provides an excellent medium for testing Watson's abilities while showing the world what it can do, but IBM has far bigger plans after "Jeopardy" than, say, booking Watson on a future episode of "Are You Smarter Than a Fifth Grader?" Rather, the company envisions Watson's underlying technology, which boils down to an efficient analytical engine capable of pulling data from multiple sources in real time, discovering patterns or relationships among data sets, and delivering an answer with a high degree of reliability, as having applications in an array of fields, from health care to efficiently running a city.