Only if you don't feed it any info from natural therapies (and alternative).
On Fri, Jan 18, 2013 at 11:55 AM, Jed Rothwell <jedrothw...@gmail.com>wrote: > Whoa! I just thought of an interesting scenario. > > That article says that the first application they intend to use Watson for > his cancer treatment recommendations. Suppose they feed Watson the entire > corpus of cancer research, and he concludes that cancer treatment has no > effect on the prognosis. Some experts believe that to be the case. Some > relevant quotes are below. The gist of the skeptical view is that diagnosis > has improved but the prognosis has not. In other words, they used to know > you had cancer five years before you died, but now they find out 10 years > before you die, so five-year survival rates have doubled. But the treatment > action does no good. > > Here are the quotes: > > "Overall, cancer mortality in the United States is unchanged in the last > 25 years and higher now than it was in 1950 (even after taking into account > the aging population) because a rise in the number of people developing > cancer has swamped any improvements in treatment. As recently as the mid > 1990s, an expert trying to measure the benefits of medical care ignored > cancer because he considered the effects of treatment negligible. ...” > > NCHS, Health, United States, 2003, p. 136 > > . . . [A] task force assembled by the public health service . . . refused > to recommend screening for lung cancer or diabetes. Even if people with > these chronic conditions go to doctors for their problems early, most will > continue to deteriorate." > > J. P. Bunker et al., "Improving Health: Measuring Effects of Medical > Care," Milbank Quarterly 77 (1994), p. 225 > > Quoted by Farley and Cohen, "Prescription for a Healthy Nation," Beacon, > 2005 > > > So imagine sometime this year they set up Watson in a medical center, and > a doctor submits a series of test results and observations of a patient. > They ask Watson, “what treatment do you recommend?” Watson says: “The > benefits of all known treatments are negligible. [CITE NCHS] The patient > will continue to deteriorate no matter what you do. [CITE BUNKER] I suggest > you let the person die in peace, rather than poking needles into him for no > reason. Human suffering without purpose is unethical. [E. WIESEL]” They > keep submitting different patient profiles, but Watson keeps coming up with > this answer. The machine has no financial or emotional investment in the > success of treatments, so it sees reality clearly. > > Now there's an interesting SF story! Coming soon to a hospital near you. > > These days a lot of things that used to be SF are finally becoming real. > > - Jed > >
