IBM supercomputer Watson is set to square off tonight in a historic Valentine’s Day encounter against Jeopardy champions Ken Jennings and Brad Rutter in a landmark test of artificial intelligence.
Competing on a televised game show may seem like a trivial challenge for a multimillion-dollar machine, but for IBM scientists the competition represents the next evolution in the quest to design computers that can beat humans at their own game. The competition lasts three nights and concludes on Wednesday.
Watson is only the latest big idea to come out of Big Blue’s elegant Thomas J. Watson Research Center in suburban New York (pictured.)
In 1997, IBM supercomputer Deep Blue defeated chess world champion Gary Kasparoff in what one observer called “the most spectacular chess event in history.”
Tonight, when Watson tries to replicate Deep Blue’s success, this time against the best Jeopardy players in the world, viewers will see is a true-blue application of artificial intelligence at work in 2011.
“It’s not science fiction anymore,” Dr. Bernard S. Meyerson, IBM’s vice president for innovation, told Wired.com during a recent interview at IBM’s Watson lab in Yorktown Heights, New York. “The human brain is an awesome thing. The reason Watson is amazing is that it took a human brain to design the actual underlying software and structure. This isn’t something that the machine invented. The machine is a consequence of it.”
At a time when many young entrepreneurs, venture capitalists and Wall Street bankers are pumping internet startup valuations sky-high, IBM’s Watson project serves as a reminder that American innovation is about more than conjuring up a deal-of-the-day website or latching onto the latest social media gaming craze.
For IBM, the Watson Jeopardy challenge is a way to make technology accessible to people through popular media. And in familiar IBM style, the company is thinking for the long term.
“Grand challenges like this move the needle,” said Meyerson, who was appointed an IBM fellow, the company’s highest technical honor, in 1992. “If you get it right, the world will be different when you’re done.”
The grand prize is $1 million; second place wins $300,000; third place receives $200,000. Jennings and Rutter have pledged 50 percent of their winnings to charity; IBM will donate all of its prize.
For the 25 IBM scientists who spent four years gearing Watson up for Alex Trebek’s scrutiny, tonight is the culmination of a labor of love. Powered by 90 IBM Power 750 servers, Watson uses 15 terabytes of RAM, 2,880 processor cores and can operate at 80 teraflops, or 80 trillion operations per second, according to the company.
In the interview, Meyerson sought to move beyond simplistic man-vs-machine debates.
“We did this because we’re up against something that’s an act of genius, the brain,” said Meyerson, who earned his Ph.D. in Solid-State Physics from the City College of the City University of New York in 1980. “The fact is that a human being is impossible to beat right now, in the sense of power efficiency, because you’ve got this little 20-watt thing, the brain, going up against many kilowatts.”
If Watson wins this week, it will be a milestone in the development of artificial intelligence.
Eric Brown, a senior researcher on the IBM team told Wired.com that Watson employs what IBM calls “deep analytics” that use natural language processing, computational linguistics, machine learning and statistical analysis to evaluate the evidence provided in each Jeopardy clue.
“We use an underlying machine learning framework to learn how to weigh all of these analytics and their scores to produce the final ranked answers and their confidences,” Brown said.
Researchers scanned some 200 million pages of content — or the equivalent of about one million books — into the system, including reference texts, movie scripts and entire encyclopedias, including Britannica, World Book and Wikipedia.
Watson “reads” the millions of pages of content in its “brain” in less than three seconds. The system is not connected to the internet, but totally self-contained.
So could Watson become “self-aware,” like Skynet, the malevolent machine from James Cameron’s Terminator series?
“People sometimes have this fearful reaction, which is nonsensical if you actually go back to the origins and think about where Watson came from,” Meyerson chuckled. “It’s not that it magically lurks and grows itself — that’s a great movie — but that isn’t what we’ve got here. This is an incredible augmentation of what we can do, and that augmentation comes from people.”
When it comes to American technology innovation over the last hundred years, IBM is unparalleled. Big Blue doesn’t generally make a big show of talking about itself — the company’s track record of achievement speaks for it — but it’s worth noting that IBM owns far more technology patents than any other American technology firm.
The Watson Research Headquarters, designed by Eero Saarinen and completed 50 years ago this April, is one of the last, if not the last, of the major historical American technology research labs still in existence and functioning, according to Meyerson. This year is also IBM’s 100th anniversary.
In designing the structure, Saarinen wanted the IBM scientists to feel close to nature, according to Jennifer Hall, a 26-year IBM Research veteran, and the in-house institutional memory of the lab. When the building was completed in 1961, Architectural Digest pointed out that it was the thinnest “skin” of black glass ever applied to a building of that size in the world, she said.
Looking west from underneath the classic Saarinen front awning, which is reminiscent of the old TWA terminal at JFK, which he also designed, one can see that Saarinen was really into airports and the concept of flight.
The front entrance is “guarded” by twin bronze sculptures inspired by the classic myth of Jason and the Argonauts. Argos 1 and Argos 2 were commissioned by Saarinen. But they might as well have been built last week to inspire the Watson challenge.
“It’s the search for the unknown,” Hall said. “Their twisting and turning forms represent the evolving nature of research. They’re perched like two giant birds on those pedestals about to take off. Research, if you will, about to take flight, coming to fruition.”
Source wants to know if we're going to hit singularity soon or if the computer will lose it all in the last round