Click here to preview the new Fast Company

Want to try out the new

If you’d like to return to the previous design, click the yellow button on the lower left corner.

David Ferrucci on Watson, the Jeopardy Supercomputer

David Ferrucci
Principal Investigator - Watson Project, IBM
Yorktown Heights, New York

The principal investigator for IBM's DeepQA/Watson project, Ferrucci, 49, led the creation of the Jeopardy-playing robot, Watson. The robot is built on data-analysis technology that could have applications in myriad sectors.

"What's really going on under the hood is that Watson is taking input, generating hypotheses, and trying to prove them based on what it thinks it has understood about the clue, the category, even the surrounding clues. For each possible answer, Watson tries to find all the material that supports or refutes that answer. Then it analyzes that evidence to come up with a set of the best answers. That's exactly what you do when you try to solve a problem, whether it's differential diagnoses in medicine or a technical-support question or financial analysis; you look for the conditions in which a particular situation might occur. What the computer is helping you do is organize all that evidence for the possible answers — the possible diagnoses — from a huge amount of content."

Add New Comment


  • Udayan Banerjee

    It is no doubt that Watson is a great achievement. However, it is disappointing to see IBM trying to use the technology to push their hardware. Just listen to Kerrie Holley (IBM Fellow looking for Watson’s next job) in the TED talk - "Experts and IBM insiders break down Watson’s Jeopardy! win" or have a look at the transcript of what he says at:

  • Fred Bothwell

    The brilliance of the Watson's team attainment in developing what is arguably the world's most advanced inference engine was unfortunately tarnished by the fact that in its Jeopardy appearance, buzzer control - Watson's unbeatable reaction time when both Watson and human competitors knew the correct response - was the basis of an ultimately unbalanced comparison of human and machine problem solving ability. As Mensa member, a one time Jeopardy contestant and the former product manager for IBM supercomputers, I'm more curious about how Watson - or a Watson-like system - might score in something like an SAT, a GRE, or the LSAT, compared to the best human (skinware) performance!