Born in the headquarters of one of the world's largest technology companies, IBM's Watson is a computer built to tackle the pernicious problem of open-question answering. While computers are excellent at performing lightning-fast searches based on keywords, computer developers have long been frustrated at the inability of artificial intelligence to properly understand context recognition and the complex relationships involved in human communication and language. Watson was designed specifically to try and defeat this long-standing problem in a unique way: developers could test its capabilities by having it play the popular American game show, Jeopardy!®.
IBM has a long history of presenting its workers and developers with “Grand Challenges,” or tasks meant to push the boundaries of technology. A previous Grand Challenge computer, Deep Blue, shocked the world in the 1990s with its capability to beat chess grand masters at their own game. The project that resulted in Watson is often cited as the follow-up to Deep Blue, though the challenge of creating a computer that could mimic natural language associations was considered to be a much harder task.
The first work on Watson began in 2005, inspired in part by the historic 74-game winning streak of Jeopardy!® champion Ken Jennings. The artificial intelligence was at first deemed by many to be too difficult to perfect, with early versions of the system taking minutes to complete questions that could be answered in mere seconds by competent human players. In addition to developing the computer to be able to recognize the context and inference conditions of a question, the IBM team also had to build Watson to be lightning fast.
In 2008, IBM began talks with Jeopardy!® executives about a competition between the computer and two previous Jeopardy!® champions, Ken Jennings and Brad Rutter. In 2011, the computer successfully defeated both champions in a two-game series. Though the computer dominated all three games, glitches in the programming caused a few comical moments. One major issue was that Watson could not hear or understand other player's wrong answers, thus occasionally giving the same wrong answer directly after a human competitor. Some critics also suggest that Watson has an unfair advantage, since it is able to press its answer buzzer faster than human beings could process the impulse to chime in.
IBM officials have expressed hope that Watson's Jeopardy!® performance is only the first step in a revolutionary form of natural language programming. After the success of the Jeopardy!® challenge, Watson is undergoing additional research and programming to extend its capabilities. Medical and legal fields have both been suggested by IBM team members as new frontiers for Watson's development.