http://www.forbes.com/sites/johnnosta/2013/03/03/is-watson-in-jeopardy-this-oncologist-thinks-so/

"Watson will likely be invaluable for integrating new medical data and published content in the medical literature. But it is simply a fool’s errand to distill a “best workup” or “best treatment” for the more complex cancer cases.  The medical recommendations it creates based on a learning process of digesting case histories from Memorial Sloan-Kettering Cancer Center (MSKCC), Cleveland Clinic, and other institutions can only be as good as the data that are input and the decision-making processes it is charged to recreate. Not to slight the excellent physicians whose views shaped the algorithms developed, but they are still the product of human minds and their own subjective interpretations of limited data, laced with significant biases.  A lung cancer patient seeing three of the best experts in lung cancer at different institutions around the US about their complex case is likely to receive at least three different opinions for a proposed treatment plan, despite the fact that all of the experts are working from the same limited set of information.  Their thoughtful commentary is infused with their own interpretation and biases."