By now many of you know that the IBM computer "Watson" blew away the competition on Jeopardy, and showed that very clever algorithms, as well as four terabytes of hard disk storage, which it could rumble through in milliseconds, can dig up trivia faster than anybody we know.
Now, we sometimes play Jeopardy at home, but it's Old People Jeopardy, which works like this: some gives an answer, such as "She said ' just put your lips together and blow." Then, I would say " I know, I know, it was, it was the woman who was in those movies we used to see at the Brattle Theater. You know with what's his name, all the time around exam time. You that was the one about, was it the statue, no it was the hurricane. It was, it was, I know, Lauren Bacall. And I would get credit for that answer.
But Watson' hard drive is cleaner than that, and the algorithm makes it's associations differently, without all of the personal history and going to the movies during final exams, and of drinking too much coffee, and being all hyped-up and happy because of going to the movie with the woman who you are eventually going to marry, of all things, which is something that Watson has no concept about how that can feel.
However, Watson is going to be moving into the health care field. In a collaboration between IBM and Nuance, the company that has vastly improved speech recognition software. They will be using it to improve the diagnostic ability of doctors.
As the press release says:
IBM offered a scenario in which a doctor who is considering a patient diagnosis could use Watson's analytics technology and Nuance's voice and clinical language understanding solutions to consider all the related texts, reference materials, prior cases, and latest knowledge in journals and medical literature. Armed with this information, Watson could help medical professionals determine the most likely diagnosis and treatment options.
Clearly, the doctor part of this team is the most expendable. The doctor could just as easily be replaced by a nurse, and then by a "technician" and then by the kid who knows more about video games than he does body parts, but he is more comfortable relating to a machine.
Now don't say that this can't happen to psychotherapy. Perhaps you are thinking that psychotherapy is too personal, and that therapists are one of the last remaining professions that utilize the powerful healing value of genuine, human understanding and empathetic interpersonal interaction. And that is true, for that is probably the basic force that psychotherapy brings to bare.
However, many years ago, back in 1966, an MIT professor Joseph Weizenbaum, who was studying linguistics devised a program on a very early computer that would simulate client-centered therapy.
Weizenbaum, J., "ELIZA -- A computer program for the study of natural language communication
between man and machine", Communications of theACM 9(1):36-45, 1966.
This program gave students the chance to sit and tell their problems to a non-judgmental,
confidential machine, and they loved it. They talked to the machine about things that they
had never talked about with real people. It seems that so many students liked to talk to the machine so much that Dr. Weizenbaum became afraid that they were becoming dependent
on it, and he had to remove their access to it.
That program was a very early version of artificial intelligence, and it has been improved
and updated and has been used recently in the Turing Test contests that have been held for over a dozen years. There is a good article in the recent Atlantic about being the human in that contest.
So, here we have Watson, and IBM and Nuance, moving into medicine. Is that bad?
Probably yes and no.
Old people will still make a big fuss about wanting to have a human doctor to hear their
complaints, but younger people will probably have more faith in robotized medicine, have
grown up spending more hours with computers than people anyway. Many young people believe that machines are better at everything anyway. Many techies are waiting for the Great Singularity, when we will all be absorbed into the cloud and become one with the cosmic, computerized, web intelligence.
It's just that another whole bunch of professionals will be out of work.
They will be home playing video games.
{note: somehow the technology of this post could not provide consistent formatting, for
some reason that some of you may understand -- but not me.}
5 comments:
Well, they didn't call it ELIZA in the past, they called it a diary.
A diary is very good for getting to the bottom of things but it can't help you out of there.
And a machine can only reply in the manner it's been programmed. It can't just use its "common sense" like a human can. Cutting comes in slot A and Drinking comes in slot B and that's the end of it.
A machine won't tell you that a vacation might be better than pills in your case, nor will it intervene in your behalf because it "knows" that you just a good person caught in a bad time.
A machine also needs electricity.
This could be a good thing, at least in the case of physical medicine. The Washington Post often carries stories of people who suffered from mysterious illnesses for months, even years. Eventually, in cases with good outcomes, someone takes a fresh look at the problem and hits on the correct diagnosis. I suppose it would depend on the skill of the programmers, but a computer might be helpful at such times.
On the other hand, when I think of all my daughter has learned from her therapist--how to challenge and reframe the thoughts that get her in trouble--I can't believe that a machine could ever be so responsive. Her therapist has greatly enhanced her quality of life; in fact, he might even have saved it.
Old people's Jeopardy...lol.
I could almost understand telling a machine your problems. Non-judgment is a big thing when you are confiding your issues. But I don't think I could imagine talking to a machine for myself.
I find it hard to believe that a machine can do therapy as effectively
as a human being can
Post a Comment