Watson, Jeopardy and morality


The Yiddish novelist Isaac Bashevis Singer once wrote that a knife could be used either to cut bread or kill. There is nothing inherently good or bad about a knife. It is simply a tool used by human beings, for better of worse.

In this way computers are like knives. For example, they have made possible the wave of democratic revolt sweeping across the Arab world. But computers also have given unprecedented power to the government and police. While it is it is good to see that the emperor often has no clothes, it isn’t so good that the emperor sees us without our clothes.

I think about this as I watched for the last couple of nights an epoch-making event, one that for me was as momentous as men landing on the moon. Here was Watson, the IBM computer, competing on the TV game show Jeopardy against two former champions. It didn’t take long to recognize the inevitable and awe to set in. That Watson was going to win was clear at the end of the first day; it was quicker to the buzzer than humans could possibly be and it had a computer’s store of facts at its circuit tips.

The amazing part, though, wasn’t Watson’s speed at spitting out facts or calculating moves. After all, the Internet gives everyone instant access to the world’s facts, and computers have been humans at chess for some time. What is new and amazing is that Watson was responding to complex speech and answering appropriately in the form of a question. Jeopardy’s questions require a facility with words that until now were believed to be a uniquely human trait.

A language-based computer is startling. It is hard to think of Watson as dumb; what it exhibited is something very close to thought.

In the near future, Watson will probably have a profound effect in the workplace, replacing many jobs with this new machine. The impact on medicine will be tremendous, I think, allowing physicians to access instantly everything that is known about particular symptoms, illnesses and cures. Whether it will create new jobs for the unskilled is a huge question. What is sure is that something new is upon us and it will force upon new sets of problems and possibilities, and it will force us to think about what it means to be human in new ways.

What Watson won’t replace is ethics. Moral choices remain human choices. But the choices the Watson will confront us with will be huge and unprecedented.

Advertisements

3 thoughts on “Watson, Jeopardy and morality

  1. Information retrieval and formulation are very important parts of medical decision making and are already available to doctors through web sites such as UpToDate, et al. Thousands of doctors comb the literature and process it into clinically useful form which is immediately available via computer during a clinical interchange. We don’t need Watson for that, although perhaps robotic thinking could refine searches.

    Medical decision making occurs on the basis of clinical input, which depends in turn on the ability to acquire accurate information historically, describe and appreciate physical findings and to integrate past medical history and test evaluations, processes which in themselves require judgement.

    When you couple information retrieval to medical decision making you produce effective medical care. I don’t think that Watson will become an effective clinician.

    Milt Masur

    • The amount of information useful to clinicians increases dramatically day-by-day. Without today’s computers, most clinicians wouldn’t know about much of the latest findings.

      Watson takes the abilities of today’s computers to be aides to clinicians to new levels. By asking questions in natural language, information will be retrievable instantly that otherwise might take days or weeks to find.

      This is a good thing, for it then gives the clinician more time to relate to the patient by helping the patient make sense of what is going on. This is where judgment is important.

  2. Hmm, this is an interesting question about whether computers can be programmed to make ethical decisions. There was an article in the Times yesterday called “The value of a life” that described how in various federal agencies, the dollar amount assigned to human life for the purposes of measuring costs incurred by, say, workplace death, is calculated based on theories about how much others in society are willing to spend to prevent the death of another individual. This number is not random, but reflects a general agreement on the numerical value of a life. These numbers, in turn, impact decision that industries make about installing safety features in cars. This quantification of human life and ethical choices (to save a life versus not) strikes me as so mechanical that a computer might do just as well, if not better, at moral reasoning. 😦

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s