The recent victory of IBM’s supercomputer, Watson, over two of the best Jeopardy game show players brings forth the question: What can Watson do for a day job? The answer that immediately crossed my mind is to become Dr. Watson. Imagine a physician who is not bound by the scope of experience, the size of his/her memory, and by the prejudice of geography. For example, someone in Massachusetts could go to the doctor with symptoms that mimic bubonic plague: swollen lymph nodes, lethargy, high fever and so on. What is the chance of the Massachusetts doctor diagnosing the real cause: Tularemia, a disease endemic to the Central Valley of California? The patient could have been bitten by a tick while visiting in Hanford for a few days. Recently in India an African disease was spotted by an Indian doctor who had spent time practicing in Africa. The patient had never left India; the disease was only now starting to show up in the country. Without the physician’s out-of-country experience, the sickness would likely have been misdiagnosed.
What Dr. Watson’s advantage would be is that every symptom ever noted could be at hand for immediate consideration. A correct diagnosis would become part of Watson’s database and an incorrect diagnosis would also be noted for later reference. One of Watson’s features is his degree of certainty (named after former IBM president Thomas J. Watson, Watson is a he). If his certainty is low, there wouldn’t be a firm diagnosis and the case could go to a live physician.
How many Dr. Watsons would we need? That depends on how fast he could handle queries. Perhaps at the start he’d get difficult cases referred from practicing physicians, then evolve into “Dial-a-Docbot” where you could call from home or access him via the Internet.
IBM has a real winner here, one that can do so much more than handle Jeopardy queries such as, “This is the best, most insightful, most humorous and brilliant blog on the Internet, ever.”
“What is Musings of a Slowly Rotting Mind?”
Update — I guessed right: IBM says Watson’s first real-world use will be to help doctors. See this article in Technology Review.