Can you hear me, Watson?
Big Blue has created a software program called "Watson" that professes to understand and synthesize plain-language queries fast enough to compete with human contestants on the quiz show.
The showdown is not yet scheduled, but plans are already underway to ship a Blue Gene supercomputer to Los Angeles for the contest, The New York Times reports. Among the potential challengers for Watson is former Jeopardy! champion Ken Jennings, who earned worldwide acclaim by winning on the show 74 consecutive times.
In that sense, Watson -- named for IBM founder Thomas J. Watson, Sr. -- is reminiscent of the Deep Blue chess-playing supercomputer that initially defeated world champion Garry Kasparov in a contested 1997 match.
The ability to understand and synthesize endless combinations of moves made Deep Blue unique. Watson will have to do better. Not only will it draw on a massive database connected to the Blue Gene supercomputer, but the software itself will also have to understand meaning and context in order to know what to search for.
Therein lies the threat to Google, though you won't hear that from IBM, Google, or Jeopardy!'s producers. "Isn't this just Google on steroids? No, it is not," Jeopardy! producer Harry Friedman said in a demonstration video posted to YouTube yesterday.
He's wrong. Watson is disruptive in its proposed alterations to the search process. In Google, users supply meaning and context, and winnow results accordingly. Can't find the answer you're looking for? Try new search terms. Watson aims to be both pervasive and precise, right from the start.
At the heart of the new system is a natural language technology that seeks to better understand questions. Do that, the creators of Watson reason, and the size of the database behind the query engine matters a lot less.
In simpler terms: If today's version of Google is a howitzer, Watson could be a high-powered sniper rifle. Both do the job, but one is far more efficient than the other.
The search for better search
Yet there are good reasons to be skeptical that Watson will disrupt Google, or Microsoft
Part of the problem is that we often erroneously link A.I. research, a software-driven process meant to teach machines to reason as we humans do, to the rise of giant supercomputers. That's understandable. The very A.I.-sounding Thinking Machines was a Massachusetts maker of supercomputers that failed in 1994. Sun Microsystems
Meanwhile, both governments and private firms continue to apply computational horsepower to the world's biggest problems as if that sheer brute force were a form of A.I. That's why IBM, Cray Research
Cray, in particular, generated $283 million in sales in 2008, according to Capital IQ, a division of Standard & Poor's.
Be better, Google
Google operates differently. Instead of investing in these sorts of "big iron" systems that Cray and IBM produce, its search engine is supported by hundreds of thousands (or perhaps a million) small servers, some of which are reportedly homemade. Algorithms capture the collective horsepower of these systems to index and present a lot of data -- the entire Web, in this case.
The problem with Google is that it's a compass, rather than a guide. Rarely can it answer a complex question as Watson promises -- it's search 1.0 to Watson's potential search 2.0.
Your move, Google.
Get your clicks with related Foolishness:
Fool contributor Tim Beyers had stock and options positions in Google and stock positions in IBM and Oracle at the time of publication. Check out Tim's portfolio holdings and Foolish writings, or connect with him on Twitter as @milehighfool. The Motley Fool is also on Twitter as @TheMotleyFool. Its disclosure policy is gearing up for a full-contact chess match shortly. Would you kindly hand it some pads?