Thursday, February 17, 2011

Watson wins at "Jeopardy"

"Computer Wins on 'Jeopardy!'" reports the New York Times in one of its lead articles today (Feb. 17, 2011). I, for one, would have been shocked if Watson the computer hadn't won. Computers, if properly algorithmically programmed, have to be superior in arriving at factual conclusions than are human beings. When a human analyzes a problem, his brain is also employing algorithmic analysis to arrive at a conclusion. Algorithms use appropriate input and go through well defined instructions to calculate results (one or more.) The brain considers all the possibilities when given certain facts and, when all the these facts are evaluated, arrives at the best possible conclusions, in the order of most probable to least probable. We put our brains though this process every day - making decisions based on past-entered input.

But a brain can not be as proficient as a computer at this function. A brain is subjected to immaterial input that may interfere with a defined algorithmic undertaking. Fatigue, the subconscious, prejudice, poor memory, and, possibly, our genetic composition, may unduly influence the defined algorithm. After all, "to err is human."

A computer is unaffected by such human traits. Computers, properly programmed, do not err. With relevant input, they will function far better - unaffected by memory issues, emotions, or prejudice. A computer's memory is essentially boundless. The retrieval of required algorithmic facts occurs at almost lightning speed. A computer can, therefore, outperform the human brain! "Jeopardy" has so demonstrated.

I am a physician. A physician's job is to obtain information from a patient using various techniques, then to input the accumulated data into his brain, and finally arrive at a diagnosis via mental algorithmic analysis. But we physicians are human! We err. We err as a consequence of these human variables.

I feel quite certain that a medically sophisticated Watson (Dr. Watson, I presume) will outscore the diagnostic acumen of the physician in the near future, and will become the physician's most valued "partner." Dr. Watson will "outdiagnose" all of us. Not only outdiagnose - but "outmanage" as well, because the management of a medical problem is also an algorithmic conclusion. We physicians won't accept Watson, at first. Like so many advances in medical management, we will have to be dragged to it "kicking and screaming," but eventually accepting it in the end.

But the human brain is still the master at innovation and new ideas in research and development.

And, we still need the flesh and blood providers of medicine to furnish, through their "art," the very important human functions of compassion and concern, which, even today, remain, arguably, the primary content of the doctor-patient interaction, and among the most crucial forms of therapy.

2 comments:

cigarman said...

Yes... but...

1) Computers don't err in following the instructions they are programmed to do. However, those rote instructions may be flawed as they are driven by a human who is not infallible.

2) Not all decisions can be deduced by a set of rote instructions. Creative thinking in problem solving is something computers will always lose.

Carl Steeg said...

Thanks cigarman - yes garbage in-garbage out. It all depends on non-flawed instruction. So does the human brain. But a properly programmed computer will not err as often as a human brain was Watson has shown us.

I certainly agree about creative thinking, as indicated in my concluding remarks.