Man versus machine: Chalk one up for the latter in 'Jeopardy!' showdown
February 17, 2011 5:00 AM
Eric Nyberg, a professor in the Language Technologies Institute at Carnegie Mellon University who worked with IBM developing the computer Watson, watches with a group on campus as Watson competes against people playing "Jeopardy!"
By Dan Majors Pittsburgh Post-Gazette
The answer is: During Wednesday evening's telecast of "Jeopardy!," they were overwhelmingly rooting for the computer to beat the humans.
The question: Who are the computer science students at Carnegie Mellon University?
More than 300 CMU students crowded into Rashid Auditorium on the Oakland campus Wednesday to watch a computer called Watson get the best of "Jeopardy!" champions Ken Jennings and Brad Rutter.
Watson earned $77,147 in the three-day competition, blowing away the $24,000 for Mr. Jennings and the $21,600 for Mr. Rutter.
But while the CMU students cheered the computer's victory -- and the scientific milestone it represents -- they still expressed sympathy with the human contestants.
"It's very revolutionary, and I think it could be a big turning point," said Dixie Kee, 19, a sophomore studying computer sciences. "I was kind of divided, rooting for both sides. I was, like, 'Watson is so cool, I hope he wins.' But I felt kind of bad for the humans about halfway through."
The gathering was part viewing party -- with free pizza, snacks and cake -- and part classroom experience as the students learned how the IBM researchers developed a giant computer that could excel at a popular question-and-answer game show format.
It also was a bit of a recruiting tool for IBM, which made sure the company got contact information for students in attendance.
"We're very proud of this, and these are the kinds of things that top research universities get involved with," said Mark Sherman, who works with IBM in Pittsburgh. "Obviously, we want to attract top talent."
CMU has a vested interest in Watson. It was the first university to sign on to the project four years ago, and CMU staff and graduate students contributed many of the advancements that made the project a reality.
"CMU was a key collaborator," Mr. Sherman said. "They were researching the same kinds of limitations [IBM] was facing in these systems. There were meetings and together they designed the new architecture."
Eric Nyberg, a professor in CMU's Language Technologies Institute, said Watson presented numerous challenges, such as designing a computer that could recognize the syntax, semantics and context of human language. Such things were elemental in developing a computer that could answer people's questions.
And, if the computer was going to play "Jeopardy!," it would have to be fast.
"The computer also had to be smart enough to know to hold its tongue if it didn't have the right answer," Dr. Nyberg said. "'Jeopardy!' was a scientific playground for us."
Indeed, Watson had several wrong answers that produced laughter in the auditorium, most notably when the computer answered "What is Toronto?" when the category was "U.S. Cities."
CMU doctoral student Nico Schlaefer, 28, who worked on the project for three years, said he was nervous when he watched Watson take on the humans.
"Something bad might happen. Watson might crash and the system might fail," he said. "It's always a legitimate concern. Watson is a very complicated system. There can always be a hardware or software problem. Complicated programs always have some bugs in them. It's unavoidable."
Mr. Schlaefer and fellow doctoral student Hideki Shima, 30, said there was a sense of accomplishment among the researchers involved, not only because Watson won a game show, but because of the practical applications the technology promises in the fields of medicine, law, business and government.
"We feel we've advanced the basic science here," Mr. Sherman said. "And that makes it a reusable skill that other people can pick up as well and build on it. It should get better and better.
"It's the nature of research. There's always more places to go."
But even Watson's fans aren't too concerned about machines beating humans on too many levels.
"We're still pretty far from actually replicating the human brain," said Michael Benisch, 28, a doctoral student in computer sciences. "And Watson is huge. The human brain isn't that big.
"Still, this is a proud moment for us, especially here at CMU. I know a lot of the people involved, I've seen them here, and it's great to know that that kind of ground-breaking stuff is happening here. The kinds of things that we're going to be able to do with this technology and the way the world is going to change. It's very exciting."