Artificial intelligence plays Jeopardy!
February 15, 2011
IBM has developed an artificial intelligence named Watson, built to parse questions in a natural (human) language and answer them.
Watson uses ideas from several areas of artificial intelligence, such as natural language processing, knowledge representation and machine learning. It has been adapted to play Jeopardy! against human contestants, and this week it is competing against two former Jeopardy! champions on CBS. The Department of Computer Science has received some local publicity in a San Angelo Standard-Times article about the event.
Each spring at ASU Rob LeGrand teaches Computer Science 4318, Artificial Intelligence, which covers some of the concepts and techniques that have made Watson possible, many of which are also useful when creating computer games. It is a required course for the Certificate of Computer Game Development. Central to the course is a series of Agent Challenges, in which students program intelligent agents to compete against (and/or cooperate with) other students on particular tasks.
You can find out more about Watson at IBM’s website.