IBM Supercomputer to Compete on Jeopardy
People interact with computers through programming languages and user interfaces, which are tools for issuing concrete commands. But when querying one another, humans use language in dense, sophisticated, and ambiguous ways. What if computers could keep up with natural language, and answer questions the way humans ask them of one another?
This is the challenge shortly to be faced by IBM's Watson - a computing system built on 'question-answering' technology called DeepQA, and implemented on IBM Blue Gene supercomputing nodes by a team at IBM Research headed by Dr. David Ferrucci. Sometime this Fall, a stand-alone Watson system will compete with top-tier human players on the long-running game show JEOPARDY!, where contestants are tasked with phrasing questions to match cleverly contrived written answers. In preparation for the real contest, the Watson system has scored well against highly-ranked JEOPARDY! players in mock shows, conducted over the past year.
This video examines what promise Watson and DeepQA hold for business, research, and society at large.