Come in! Come in!

"If you are a dreamer, come in. If you are a dreamer, a wisher, a liar, a Hope-er, a Pray-er, a Magic Bean buyer; if you're a pretender, come sit by my fire. For we have some flax-golden tales to spin. Come in! Come in!" -- Shel Silverstein

Thursday, February 17, 2011

Watson

(Fade in: The familiar musical theme from Jeopardy. The voice of Don Pardeau announces): "It's time to play Jeopardy!" (Fade out music. Audience applauds).

"I'll take 'What's on TV?' for $200, Alex."

"Answer: 'The quality being tested by an IBM Computer and two human contestants."

"Question: 'What is intelligence?'"

That answer-in-the-form-of-a-question is not simply the entertainment value of the last three nights of the game show Jeopardy in which Watson - a computer the size of 10 refrigerators, named for after IBM's founder - competed against two human contestants.

It's also The Question which has been the point of scientific and sociological and cultural inquiry for years. I suspect that question will continue to be asked for years to come.

Watson won, of course. All three nights. Although, he did miss the Final Jeopardy question the second night. "Toronto" is decidedly NOT a U.S. city, but Watson didn't understand the question because of the way it was worded.

I hate it when that happens, don't you?

As Dave Ferrucci of IBM Research says, this points out the limitations of relying so heavily on statistical reasoning. The result, as in Watson's case, is akin to a human autistic savant.

You know. Like the character Charlie Babbitt in one of my all-time favorite movies "Rain Man".

"Jeopardy!" is challenging because the clues are esoteric. No one would watch a TV program about a battle of common sense.

Hence, the popularity of "Reality Shows" like The Jersey Shore (don't get me started).

Watson is only a "statistical brain," and not an analytical one, so a question like "If a snowman melts and later refreezes, does it turn back into a snowman?" would be nearly impossible for a statistical reasoning program to tackle.

You can't answer it by calculating how many times "snowman" appears next to "melt" and "refreeze" in every article and book ever written.

Ms. Conroy reported that, in one interview about Watson she listened to, it was suggested that Watson could be used in medicine to cross reference a patient's symptoms with compendiums of digitized medical journals and data - a sort of statistical second opinion to a doctor's professional experience.

Pretty soon, the scientist said, computers will be used to more accurately diagnose and even cure diseases. Ultimately, perhaps, computers might even 'discover' a cure for everything from high blood pressure and diabetes to cancer.

Ms. Conroy said, "When that happens, you have to know that the pharmaceutical companies will get involved."

"Why?" I asked, sounding every bit as befuddled as I was.

She sighed a weary sigh in that way she sometimes does when she's talking to someone who obviously isn't in the medical profession and doesn't read as many mystery novels as she does.

"Right now, as long as we treat diseases symptomatically, the pharmaceutical companies make millions - billions - trillions! We put people on medications to control cholesterol and, once they come off the medication, their cholesterol levels sometimes get worse than they were before."

"The real treatment goal of pharmacology, then, is not to find a cure for high cholesterol - even though it can be life threatening - but to continue to treat the symptoms."

"Ah," I said.

Then, she added in that wonderful way of her ability to formulate and articulate pithy bits of wisdom, "There are great profits to be made in pathology."

You have to admit, she has a point.  Perhaps this is Phase II of the real reform of health care.

Is artificial intelligence a threat or a benefit?

I'm fascinated by the overtones of fear I hear in conversations about Watson. Mostly, they have to do with computers and machinery taking over the work of humans.

Well, guess what? That's what technology does. Tractors, forklifts, word-processing software - they all took away jobs.

And people, with their creative minds, have used them throughout history to figure out where the next jobs will be - even using the very technology designed to replace human labor to make greater profits.

At the end of the day, that's the real difference between 'artificial' and 'real' intelligence.

"I'll take "Cultural Anxiety" for $2,000, Alex."

(Bell sounds repeatedly. Audience applauds.) "And. . . . it's the Daily Double!"

"I'll wager $2,000, Alex"

"Answer: Dr. Spock, the half Vulcan, half human fictional character of the TV and movie series, 'Star Wars' said these words as a Vulcan blessing."

"What is, "Live long and prosper."

"And you win the Daily Double" (Bell sounds. Audience applause.)

That's something no computer can achieve without human assistance.

However, with the assistance of 'artificial intelligence', humans may be able to achieve both.

Peace, and long life, Dr. Spock.

I'll leave you with a little "Think Music" to consider these things.

10 comments:

Kirkepiscatoid said...

Ms. Conroy is right.

I don't make a dime off of wellness.

This might be why, after 20 years of making a living off the back of disease, I have grown more and more excited about the proposition that being a force for spiritual wellness might be what God has in mind for me.

Elizabeth Kaeton said...

Mebe so, Doc. Mebe so.

Paul said...

This is some of the best analysis of Watson I have read. Bravo.

Reminds me of a demo I heard in a graduate seminar many years back. They were demonstrating computerized speech synthesis, and they played a recording of a computer "singing" the Queen of the Night aria from Mozart's Magic Flute. I was impressed, until I realized why they had picked that particular aria.

It had no consonants. They hadn't figured out how to do consonants yet, so they picked the one aria in the entire operatic repertoire which didn't have any.

Tech demos pick very carefully the things they can do. If you don't see it done, they can't do it.

Speech recognition and natural language processing are among the hardest problems in computer science. Many Ph.D. theses have crashed and burned on this topic. If you want to demonstrate a program's limitations, try a bit of irony, sarcasm, or anything remotely poetic or metaphorical. These programs are limited to simple declarative sentences.

The demo was impressive, to be sure. But it also illustrated the limitations of these techniques.

IT said...

The real finding is that a human with a computer can run circles around watson. Combine our sort of intelligence with the iterative and exhuastive capabilities of a computer and that's where things happen.

Elizabeth Kaeton said...

Paul - Thanks. I love what you've said. Thanks for leaving your comments.

Elizabeth Kaeton said...

What a wonderful insight, IT. Thanks

textjunkie said...

Answer: Dr. Spock, the half Vulcan, half human fictional character of the TV and movie series, 'Star Wars' said these words as a Vulcan blessing."



Well, that would be hard for a human to answer, given Spock was never on Star Wars.

:)

But other than that, yeah. Not exactly the end of humanity, though Watson is an impressive display. I'm more intrigued by the engineering--good grief, the amoutn of computing power it took to just play Jeopardy! Yikes!!

Elizabeth Kaeton said...

textjunkie - Oops. Not Star Wars. Star Trek. Yikes. How did I miss that? Thanks for pointing it out.

MarkBrunson said...

Frankly, I missed it, too.

People on the computer end of Trekkism (covering both the Trekkies and Trekkers), tend to forget that Spock also said, "Logic is the beginning of wisdom, not it's end."

Computers only have logic, and, unless we can actually develop that human spark of personality in them (Asimov's positronic brain?) logic is all they will have.

Humans should use data to arrive at wisdom, not mere answers - that's just more data.

Elizabeth Kaeton said...

Thanks, Mark. I had forgotten those words of wisdom from Mr. Spock.