DXPG

Total Pageviews

Monday, May 6, 2013

David Ferrucci: Life After Watson

To the degree there was a human face of Watson, the “Jeopardy!” computer champion, it was David Ferrucci. He was the I.B.M. researcher who led the development of Watson, an artificial intelligence engine. The goateed computer scientist was always articulate and at ease in front of a camera or a microphone.

Dr. Ferrucci has left I.B.M. to join the giant hedge fund, Bridgewater Associates. And the weight of the Watson-related fame, it seems, played a role. “I was so linked to the Watson achievement, and where I.B.M. was taking it, that I felt I was almost losing my identity,” he said in a recent interview.

After Watson beat the best human Jeopardy champions in 2011, its artificial intelligence technology was directed toward new challenges, like assisting doctors in making diagnoses in a research project at the Cleveland Clinic.

Dr. Ferrucci led that next-generation Watson research as well. But he went to Bridgewater at the end of last year. Bridgewater, a private company, made no announcement of its new hire. Yet word of Dr. Ferrucci’s departure from I.B.M. has been circulating among scientists in the artificial intelligence field. And I caught up with him recently for an interview, supplemented by a lengthy e-mail he titled, “My Reflections.”

Dr. Ferrucci, 51, said he had “a great, great career” at I.B.M., spanning 20 years, and “they paid me very well.”
He said he “never imagined myself at a hedge fund,” but eventually the appeal of working in a smaller environment in an entirely new field for him â€" applying artificial intelligence to macroeconomic modeling â€" won him over.

Dr. Ferrucci said the more recent work he was doing at I.B.M., called WatsonPaths, was the direction he thought artificial intelligence research needed to go to make further advances, and it was the approach he saw Bridgewater pursuing to economic modeling.

Much of artificial intelligence today, he said, focuses on mining vast amounts of data to make predictions. Those predictions are based on statistical probabilities and patterns â€" a certain symptom is highly correlated with a certain disease, for example.

“But in a purely data-driven approach, I can explain my decisions,” Dr. Ferrucci said. “People are so enamored with the data-driven approach that they believe correlation is sufficient.”

The Big Data formula, he noted, has proved to be “incredibly powerful” for tasks like natural-language processing â€" a central technology behind Google search, for instance.

WatsonPaths, by contrast, builds step-by-step graphs, or paths, that trace possible causes rather than mere statistical correlations. In the case of medicine at the Cleveland Clinic project, for example, the paths go from an observation of symptoms to a conclusion about the diagnosis of a disease and treatment.

That approach is a hybrid of the Big Data tools, which sift through troves of medical literature, and logic tools to identify likely chains of inference â€" what humans see as logical explanations for the “why” of things. The approach is also a step in the direction of classic artificial intelligence, which relied on knowledge rules and relationships, to create so-called expert systems. The blend combines elements of what Dr. Ferrucci termed “my 30-year journey in A.I.”

At Bridgewater, Dr. Ferrucci sees a similar path to modeling the economy and markets. “Their approach to investment,” he wrote in his e-mail, “is based on a fundamental understanding of how the global ‘economic machine’ works.”

Its models, he added, are “informed by but not blindly driven by the data.” The opportunity, Dr. Ferrucci wrote, is to build “predictive systems that fit perfectly with my interests. How cool is it to imagine a machine that can combine deductive and inductive processes to develop, apply, refine and explain a fundamental economic theory?”