‘Jeopardy,’ IBM Had Reason to Be Scared of Watson Match: Books
Plenty of observers have weighed in on Watson, the computer that International Business Machines Corp. built and programmed to play the quiz show “Jeopardy!” Few have done it better than Stephen Baker, author of “Final Jeopardy: Man vs. Machine and the Quest to Know Everything.”
Baker, formerly the senior technology writer for Businessweek, got behind the scenes at Armonk, New York-based IBM to watch a team of scientists and engineers create a machine to compete in Sony Corp.’s beloved half-hour nerd-fest of answers and questions hosted by Alex Trebek.
Adding to the challenge was one of the computer’s flesh- and-blood opponents: Ken Jennings, the Joe DiMaggio of “Jeopardy!” who won a record 74 straight matches.
David Ferrucci, the chief scientist on the team that developed Watson -- named for IBM’s founder -- understood that no matter how fast the machine was, or how many facts they crammed into its database, humans like Jennings still possessed skills no one had been able to engineer with much success.
“Any ‘Jeopardy’ machine they built would struggle mightily to master language and common sense -- areas that come as naturally to humans as breathing,” Baker writes. “On the positive side, it wouldn’t suffer from nerves.”
Baker goes easy on the hard science behind Watson, referring readers to scholarly journals for technological details. Even his description of the hardware makes the technical tangible:
“The eight towers, each the size of a restaurant refrigerator, carried scores of computers on horizontal shelves, each about as big as a pizza box. The towers were tilted, like the one in Pisa, giving them more surface area for cooling.”
Instead, he plays up the skirmishes that break out at the border between person and processor. The most engaging chapter focuses on the controversy the Watson project sparked in the artificial-intelligence community.
Some scientists feared Watson would draw attention, and funding, away from their efforts to create machines that mimic human thought, a complex and not fully understood process.
“The world would see, and perhaps fall in love with, a machine that only simulated intelligence,” Baker writes. “The machine was too dumb, too ignorant, too famous, and too rich. (In that sense, IBM’s computer resembled lots of other television stars.)”
On a more mundane level, IBM and the producers of the show were concerned about image. IBM’s Deep Blue had triumphed over chess master Garry Kasparov in 1997, but chess is not “Jeopardy!” and chess tournaments in the U.S. don’t attract nine million viewers a night. Any failure by Watson could damage the brand.
“Jeopardy!” had its own concerns. The producers couldn’t be seen as rigging the game for or against the machine --evoking the specter of the quiz-show scandals of the 1950s --but they couldn’t allow the lightning-quick Watson to buzz in on every clue, steamrolling his human rivals. The solution: Watson got a mechanical thumb.
The tournament itself was taped in January and aired Feb. 14, 15 and 16. Watson’s strengths and flaws were manifest. Puns and wordplay stumped it (including a category devoted, ironically, to words found on a computer keyboard); while it excelled at more straightforward trivia such as Beatles lyrics.
Baker’s last chapter gives details of the broadcast not apparent to the TV audience, including judgment calls, equipment malfunctions (by the game board, not Watson) and the reason Watson answered a question about U.S. cities with “Toronto.”
In the end, Watson wiped the floor with his opponents, Jennings and Brad Rutter. Not only because he was knowledgeable -- both his opponents were, too -- but also because he was, after all, a machine: “It was its buzzer that killed us,” Rutter said.
Andrew Dunn writes for Bloomberg News. The opinions expressed are his own.)
To contact the reporter on the story: Andrew Dunn in New York at firstname.lastname@example.org
To contact the editor responsible for this story: Manuela Hoelterhoff at email@example.com
Bloomberg moderates all comments. Comments that are abusive or off-topic will not be posted to the site. Excessively long comments may be moderated as well. Bloomberg cannot facilitate requests to remove comments or explain individual moderation decisions.