Stephen Baker

After ‘Jeopardy’

The next job for computer Watson — hired hand

(David Gothard for The Boston Globe)
By Stephen Baker
February 15, 2011

E-mail this article

Invalid E-mail address
Invalid E-mail address

Sending your article

Your article has been sent.

Text size +

IN A man-vs.-machine match up, does the computer always play the villain? IBM’s branding experts grappled with the issue a couple of years ago when they were deciding what sort of face and voice they should provide for the computer, Watson, which is competing this week on “Jeopardy!’’ The objective was to kindle warm feelings for the bionic quiz-show player — and even entice the public to root for it. They agreed they should distance Watson from the murderous computer, HAL, on “2001: A Space Odyssey.’’ That could be a PR disaster. And it shouldn’t resemble a person. Viewers might find that creepy — a reminder that Watson-like contraptions could be angling for their jobs. No, IBM’s computer would speak with a friendly, upbeat voice, flavored with a hint of the bionic, and its face would remain an abstract globe — with no eyes, nose, or mouth.

And yet Watson still stirs fears and resentment. As the computer headed into its “Jeopardy!’’ showdown last night against human champs Ken Jennings and Brad Rutter, blogs and Twitter brimmed with fears about this next step in artificial intelligence. “I put the date at Monday, February 14, 2011. When Watson (SkyNet) goes live and crushes human intelligence on Jeopardy!’’ wrote one Twitter user. Meanwhile, a number of scientists in artificial intelligence, including several at MIT, the cradle of AI, view Watson as a cleverly-programmed dummy masquerading as a genius. Even as it masters a game of words and knowledge, they argue, it doesn’t “know’’ what it’s talking about.

Watson would not likely be the target of such criticism if it weren’t built for TV stardom. The game positions it in the role of a human. As such, it poses a threat for some, an insult for others. But the entire TV match is a contrivance, one designed to shine a global spotlight on IBM’s technology and burnish its brand. To create the drama and spectacle, IBM’s machine has to take on humans and imitate our brain functions. This positions it as an enemy and a pretender.

But the true implications of Watson’s technology will come after it retires from the stage and pursues a workaday career in offices and labs. That’s when Watson will shed its avatar and revert to its true nature, that of a powerful machine working for us, not against us. Watson will be a tool.

You might see Watson making laughable mistakes on its national turn, and wonder what value such a tool would have. In practice games, for example, it answered that a butterfly’s diet was “kosher,’’ and in one 19th-century literature clue it mistook Dickens’ Oliver Twist for the ’90s techno band, the Pet Shop Boys.

Yet there’s plenty of opportunity in the workplace for a fallible machine that (usually) makes sense of complex human language. The networked world is flowing with what computer scientists call “unstructured data,’’ much of it in words. It’s far too much for humans to read. Last year, for example, some 50,000 academic studies were published on neuroscience. Unlike a search engine, which ferrets out words and phrases from such studies, a language-savvy machine like Watson can dig through mountains of documents and come up with lists of possible answers. This service could prove valuable in call centers, police work, pharmaceutical labs, and even hospitals.

The key to understanding Watson’s value is to regard its answers as suggestions, or hypotheses. Consider Watson as a research assistant on a medical diagnostic team. A patient comes in with a puzzling set of symptoms. Watson launches a search through hundreds of thousands of journal articles and case studies. It returns with six possible diagnoses and its level of confidence in each one — along with links to the evidence it studied. Let’s say two of those six are far-fetched, the medical equivalent of its kosher butterflies. Doctors know enough to rule out a few others. Still, if even one of those six possibilities leads the team toward plausible answers they hadn’t considered, the machine will have done its job.

There’s no guarantee that Watson’s technology will provide the question-answering technology for all the industries IBM has targeted. Maybe competing companies or IBM itself will come up with cheaper, smarter, or more flexible systems. But a new generation of language-savvy machinery will soon be hunting down answers for us. Some will be wrong, some stupid. We human beings, exercising our still-unparalleled brainpower, will be the judges of that. Once you regard this technology as a powerful supplement to human cognition — and not a replacement — Watson suddenly starts to look much friendlier.

Stephen Baker is author of the new “Final Jeopardy — Man vs. Machine and the Quest to Know Everything.’’