IBM researchers have worked for two years to create a system that can understand and respond to open-ended, natural-language questions
In something that sounds like it was cooked up by the writers at “Saturday Night Live,” IBM is building a computer that will compete with human contestants on the US game show “Jeopardy.”
IBM researchers have spent two years creating a QA (question answering) system—code-named “Watson”—that they say will be a significant step forward in giving businesses the ability to quickly find the information they need.
The Jeopardy arena will be a key test for the system, given that it will have to be able to handle questions that have subtle meanings, riddles, puns and the like, then sort through the information in its database, to come up with the answer within a second or two in order to beat its human counterparts, the researchers said.
“The essence of making decisions is recognizing patterns in vast amounts of data, sorting through choices and options, and responding quickly and accurately,” IBM President and CEO Sam Palmisano said in a statement. “With advanced computing power and deep analytics, we can infuse business and societal systems with intelligence.”
IBM already has some experience in this computer-vs.-human field. In 1997, an IBM computer called Deep Blue beat chess champion Garry Kasparov. IBM researchers said Deep Blue could calculate 200 million chess moves per second based on a fixed problem.
However, that differs from what IBM is trying to do with Watson, which is to create a system powered by dynamic and intelligent software that can handle open-ended questions as normally presented by humans. That means Watson—using massively parallel processing to instantly understand complex questions—will have to figure out the subtext to the question, what the person is actually looking for, and then find the answer.
“The challenge is to build a system that, unlike systems before it, can rival the human mind’s ability to determine precise answers to natural-language questions and to compute accurate confidences in the answers,” David Ferrucci, the leader of IBM’s Watson team, said in a statement. “This confidence processing ability is key. It greatly distinguishes the IBM approach from conventional search, and is critical to implementing useful business applications of [question answering].”
And for Jeopardy, Watson will have to give the answer in the form of a question.