Tonight: Watson, the Jeopardy-Playing Computer, Faces Its Human Foes

By Andrew Moseman | February 14, 2011 1:32 pm

NOTE: Before tonight’s big match begins, check out our feature, “Who’s Smarter, a Human or a Computer? Round 9: Jeopardy,” on the other human games that AI programmers have tried to perfect—and the ones where humans maintain the advantage.

I can already hear the Jeopardy theme music (which isn’t my ringtone, I swear!). Tonight, one of the highest-profile man versus machine contests in years begins, as Jeopardy will air the matches pitting former flesh-and-blood champions Ken Jennings and Brad Rutter against Watson, the question-and-answer supercomputer by IBM.

Since traveling to IBM’s research center for the practice/demonstration match (which Watson led when play stopped after 15 clues), we here at DISCOVER have been simultaneously excited for the match and anxious about the prospects for our species’ chosen representatives to come out on top. Jennings apparently feels the same way:

“It’s a new experience for me to feel like an underdog, playing against this unstoppable supercomputer,” he says. “At IBM’s research lab, the center of the stage had a big Watson logo, like you’re playing a basketball team on its own court. I knew this was gonna be an away game for humanity.” [Los Angeles Times]

The emotionless Watson is prepared, too. IBM’s David Ferrucci and his team spent four years building Watson to understand the puns and riddles of the English language as used on Jeopardy—and pumping its databases full of more information than a human could hope to remember.

They scanned a universe of knowledge into its capacious 15 trillion-byte memory: great literature, mathematical and scientific formulas, the name of every pope and Best Actress Oscar winner. To call this compendium of information encyclopedic would do it a disservice. It’s practically Wikipedic – but without the looming threat of inaccuracy. [Washington Post]

Hopefully, for humanity’s pride, Watson is slow on the buzzer. But based on what we’ve seen so far, don’t count on it.

Related Content:
80beats: Jeopardy-Playing Computer Tromps Human Players in Practice Round
80beats: Your Clue Is: “This Robot Will Attempt to Crush Humans in ‘Jeopardy!’”
80beats: Watson, an IBM Supercomputer, Could Be the Next “Jeopardy!” Champion
Discoblog: First Chess, Now Poker? Computer Programmers Try To Crush Human Competitors
DISCOVER: Deeper Blue?

  • JaberwokWSA

    When I (and I assume most) play Jeopardy, I will typically know the answer before the clue is completely read and they throw the switch to allow the competitors to buzz in. So, in the real sense of competition, it isn’t necessarily who can figure out the answers fastest, but who can buzz in with the answer fastest. The computer will of course have faster reaction time to buzz in with an answer, so the entire concept of human versus computer Jeopardy is skewed in favor of the computer.

    I do think that the exercise of whether you can program a computer to learn to understand the subtlies of English grammar and understand the relevant bits of information is a remarkable advancement. This is the important part – not what a computer knows or how fast it is.

    But I think the concept of a computer versus human Jeopardy game demonstrates only the speed of a computer over a human. After all, give me an encyclopedia as vast to reference, and give me time to perform searches at an equivalant reaction time scaling, and I bet I can come up with answers too.

  • Daniel J. Andrews

    How is Watson given the clue? Does it register it verbally, or is it transmitted via some other means? And if transmitted, then when? As Jaberwok points out, most people playing know the answer before the clue is read. If Watson needs to hear the clue first, this gives humans a bit of an advantage (assuming buzzers are active–maybe they aren’t active until Alex finishes reading, in which case, advantage greatly lessened).

    I’m hoping we’ll see lots of word play. It’ll be interesting to see how the computer handles some of those creative puns and clues.

  • JaberwokWSA

    My understanding of the buzzer is that there is a person sitting offstage. When Alex finishes reading the clue, he flipps the switch to activate the buzzer. I believe Ken Jennings himself mentioned that the key to winning was watching that guy and timing when to buzz in. If you’ve noticed, Alex is never interrupted from reading the clue.

  • Ken

    The English language is rife with ‘word-play’ sound-alike words, made-up words puns, idioms and the rest. I made a list of idioms – at last count there were over 16,000. If I could write the answers/questions for the game, I’m sure I could stump the computer. For example; ‘the workers raised the house’ or ‘the workers razed the house’. Both sentences sound exactly the same, yet mean opposite things – so it’s added context which reveals what is being said. Also, emphasis on certain words can change the meaning of a statement. Probably most confusing for a computer would be made-up words that a person could de-code, but a computer probably couldn’t. Additionally, there are myriad cultural twists and turns that a computer couldn’t handle. Example; which Beatle had the Beatliest mop of the bunch? A person could decipher that oddball question, but it’s doubtful a gazillion bit computer could.

  • nick

    1) Watson does not buzz in until he is confident of an answer (programmed-in behavior) – something that humans are not limited by, and in fact Jeopardy! champions say they routinely buzz in before they’re sure they know the answer and use the extra time to think up the answer. Ken Jennings did indeed say this is part of his strategy (as well as watching the guy activate as JaberwokWSA mentions).

    2) It’s not just looking up answers in an encyclopedia – as Ken sorta mentioned, Jeopardy! is rife with puns and word-play, something computers are f-cking stupid at. That’s why Watson is such a breakthrough in machine intelligence. This is actually the main challenge and accomplishment of the Watson research team, understanding the vagaries of English and dealing with the clever mis-direction of puns and other word play.

    3) Watson is fed the questions as text. They could have gone with voice recognition but A) The contestants read the clue before Alex is finished speaking, he’s mainly there to clarify pronunciations and B) if they were doing voice recognition only Watson would be hampered by homonyms and puns, as he wouldn’t be able to read the text like the human contestants could.

    4) It’s mostly a publicity stunt for IBM to flex and floss just how amazing the Watson system is – and according to reports on round 2, Watson whomped on the champs after tying one (not Ken) in round 1.

    5) I can’t wait until Google gets ahold of a Watson system and feeds in its collection of books, scholarly articles, and the cache of the whole internet (oh yes they have it) and allow us to start asking questions.


Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!


80beats is DISCOVER's news aggregator, weaving together the choicest tidbits from the best articles covering the day's most compelling topics.

See More

Collapse bottom bar