Monday, February 21, 2011

What "Watson" means


Being a geek, my favorite game show is Jeopardy!  We gather around from time to time at the day job and read Jeopardy! questions to one another, training for that shining day where I might actually get that callback to be on the show.   I try to keep myself mentally fit, reading facts from The New York Times and memorizing the originator of the "world's silliest bottom burp."  The answer to the latter is "Rik" for those of you fans of BBC's The Young Ones.

So in light of both my fandom of Jeopardy! and my fascination with cybernetics and transhumanism, you can imagine how interesting the past week has been.  An IBM computer system named "Watson" appeared on the game show and managed to best the game's greatest players.  This was quite a test for artificial intelligence.  After all, many questions on Jeopardy! deal with the nuances of language.  Would "Watson" be able to parse these subtleties?  The answer appears to be a resounding "yes."

Dr. Ray Kurzweil, author of The Singularity Is Near and a transhuman guru, wrote an essay called "The Significance of Watson," reflecting on the AI's triumph.  One of the interesting points that Kurzweil brings up is the defensive nature that people tend to take on when confronted with this form of technological achievement.  When IBM's Deep Blue computer defeated champion Gary Kasparov in chess back in 1997, many wrote it off as "well, that's just chess."  In other words, the game is not exactly exemplary of human thinking.  That position is tenuous at best, but it was there.  Now comes this victory for a "thinking machine" that seems to handle the random intricacies of human language.  Again, I am thinking people will downplay this by saying, "it's just a game show, for cryin' out loud."
Not so fast.

Kurzweil has predicted that computers will not only achieve intelligence but will attain human-like consciousness by 2030.  Watson is another brick in the road towards that point, that "singularity" wherein the difference between human and machine becomes paper thin.  This is no small achievement and I would argue that anyone attempting to downplay it is reacting in part out of fear of some sci-fi scenario where humans lose legitimacy.  If anything, artificial intelligence such as Watson will greatly benefit us.
I recently read of Watson's relevancy explained via this scenario: Imagine a patient goes into a doctor's office complaining of certain symptoms.  An AI takes these symptoms and returns six different diagnoses.  Of those six, it is likely that one or two of them will be unlikely or just outlandish.  Discounting them, if there is but one other likely diagnosis in the bunch that the doctor might not have otherwise considered, then the AI has paid for itself.  Still, we're a bit of a ways from something like Watson fitting on either our desktops or in our smart phones.  Watson runs on 90 servers.  But that will not be an obstacle for much longer.
 
This is not something be feared, but hotly anticipated.  After all, can you imagine your life without search engines?  They do what humans cannot do.  Merging that quality with the facets that humans inherently have will create an amalgamation of human and AI that will ultimately be beneficial to us all.  
Except for the fact that I may never win on Jeopardy! that is.


Follow me on Twitter: @Jntweets

No comments:

Post a Comment