Count computer scientists Bruce Porter, Ray Mooney and Ken Barker among those cheering for the machine in the Jeopardy! Challenge, which pits two human Jeopardy! champions against Watson, a computer built by IBM Corp.
Watson will take on Ken Jennings, who had the show’s longest winning streak, and Brad Rutter, its all-time money winner, in games that will broadcast Feb. 14, 15 and 16.
IBM dreamed up the Jeopardy! contest to demonstrate the capabilities of computers to do things people do. In a previous human vs. computer match, IBM’s Deep Blue computer beat chess champion Garry Kasparov in 1997.
Actually, Porter, Mooney and Barker, members of the Department of Computer Science at The University of Texas at Austin, are cheering more for the people behind Watson than they are for the computer.
“I’m not rooting for this program, which is just a piece of software,” said Mooney, a professor in the Department of Computer Science. “I’m rooting for the team at IBM that put this together. They put their heart into this.”
IBM counts Porter, Mooney and Barker as members of that team.
Although they don’t know whether Watson runs a line of their computer code, IBM credited them and researchers at seven other universities with making contributions to the computer.
The three computer scientists work on getting computers to develop common sense knowledge and understand and process human language. Porter, chairman of the department, and Mooney have worked in artificial intelligence for more than 20 years, most of them at the university. Barker, a research scientist, has been at the university for 11 years.
Mooney’s line is that the three biggest challenges to a computer being able to build knowledge from data are, “Ambiguity, ambiguity, ambiguity.”
Examples: When is a bat a Louisville slugger and when is it a flying mammal? When Annie Leibovitz shoots someone will the result be a photo in Vanity Fair or a jail sentence?
People understand the differences because we’ve built a bank of knowledge and experience since birth. Computer scientists have to program that kind of common sense context into computers.
Computer scientists have been doing it for decades. They’re making progress and Watson is part of that progress.
Watson has about 10 million documents — including Wikipedia, the Internet Movie Data Base, the New York Times archive — stored in its memory. Its task is to know what to look for, where to look for it and know how to put that data together into an accurate answer.
Jeopardy! is a tricky challenge for a computer. Players are given clues, which take advantage of the ambiguity in human language with puns and word play. It can take human contestants a couple of tries to home in on just what kind of answer the game is seeking.
Not only that, but answers to the clues must be given in the form of a question.
Porter, Mooney and Barker are excited about the Watson project and have had fun watching it unfold.
“We’re excited for a couple of reasons,” Barker said, “One because we were skeptical that it could be done. Second, it’s an entertaining task. It’s fun for us.”
Mooney said it’s good for science to have events like the challenge that capture the public imagination.
“The folks at IBM had good intuition about this,” he said.
They said that what is learned in building Watson can be applied to areas such as where the information in a field such as medicine, finance and military intelligence overwhelms the human capability to keep up.
“Watson would allow a doctor the ability to ask pointed questions about a body of literature without the doctor have to read it all,” Porter said.
Attend a Jeopardy! Challenge Watch Party
A watch party will be held Feb. 14, 15 and 16 in the Avaya Auditorium in the ACES building. On Feb. 14, James Fan, a member of the IBM Watson team and who received his Ph.D. from the university, will be at the viewing and lead a discussion about Watson.
IBM has posted videos with comments from Ray Mooney and Ken Barker. They are:
Read more about Watson and the Jeopardy! Challenge on the Further Findings research blog: