In 2011, Ken Jennings, the contestant famous for winning 74 games in a row on the television quiz show Jeopardy! played against a computer. The IBM machine, called Watson, named after IBM’s founder, Thomas J. Watson, is actually a question answering machine. This artificial intelligence comes complete with software and technologies that allowed it to understand and answer encyclopedic questions posed in a natural language. The researchers working on Watson needed to program the machine to not only answer the question, but wager bets (in something called a “daily double”), and even make educated bets on the final round.

In the end, Watson beat not only Jennings at Jeopardy, but a third player, Brad Rutter, with the final tally at $77,147 to Watson, $24,000 to Jennings, and $21,600 to Rutter. But the researchers also admitted that the machine may have benefited from what is called the “buzzer factor”—humans anticipate the buzzer to answer a question, and if they buzz too early, they can be locked out of the ability to answer by about a quarter of a second. Watson could hit the button in as little as 10 milliseconds, a time difficult for a human to beat, and in the second round Watson beat the others to the button 24 out of 30 times.

Overall, it was a victory for the field of artificial intelligence (called “the study and design of intelligent agents”), a computer science subfield that has had its share of failures over the past decades. Watson’s calculating abilities will not go unused: The researchers at IBM are now using the machine’s skills to develop a physician’s assistant service that will allow doctors to call on a “cybernetic assistant” and will eventually add voice recognition software to the service.

The architecture of the decision tree used by IBM’s Watson computer to reach conclusions based on inputted data.