The globe-like avatar representing IBM Watson is based on the IBM Smarter Planet icon. In the display employed for the Jeopardy! contest, colors and motions changed depending on game states and answer confidence levels.

2011

WATSON ON JEOPARDY!

Ken Jennings (b. 1974), the world champion of the game show Jeopardy!^®, wrote about his contest with an AI entity called Watson: “When I was selected as one of the two human players to be pitted against IBM’s ‘Watson’ supercomputer in a special man-vs.-machine Jeopardy! exhibition match, I felt honored, even heroic. I envisioned myself as the Great Carbon-Based Hope against a new generation of thinking machines. . . .”

Watson was a question-answering computer system that used natural-language processing, machine learning, information retrieval, and more to defeat the world champion in 2011 in a game involving general knowledge clues. What made the task particularly difficult—more difficult than chess playing—is that the computing system needed to provide an answer in just a few seconds, while considering the challenges and ambiguities of the English language, with clues that included puns, humor, riddles, cultural references, special contexts, and rhymes—something that humans consider instinctively.

To accomplish this task, Watson employed thousands of parallel processing units called cores, along with information such as the entire Wikipedia corpus, stored in its RAM memory (because access to spinning hard drives would be too slow during the competition). All information had to be stored locally, since Watson was not allowed access to the Internet during a match; in order to arrive at an answer, the AI considered the results of numerous separate analysis algorithms at once. The more algorithms that found the same answer, the more likely it would be a correct answer. Watson continually scored different answers with a confidence level, and if the confidence was sufficiently high, Watson would provide the answer.

After his loss, Jennings wrote “But there’s no shame in losing to silicon. . . . After all, I don’t have 2,880 processor cores and 15 terabytes of reference works at my disposal—nor can I buzz in with perfect timing whenever I know an answer. My puny human brain, just a few bucks worth of water, salts, and proteins, hung in there just fine against a jillion-dollar supercomputer.”