Tuesday, February 15, 2011

I'm sorry, Dave. I'm afraid I can do that.

Maybe I'm a day or two behind (it aired on Monday in the US, I think), but watching the first part of the Jeopardy-IBM challenge was as exciting for me as I imagine the Apollo 11 moon landing must have been back then. I saw it on youtube with some friends (who didn't know about it before), and it was interesting to see that they were pretty 'meh' about the whole thing... for them, Watson just seemed to be a glorified encyclopedia, and I don't want to blame them for thinking so. (Actually, I did blame them, but that was just my sometimes bad temper.)

Yet to anybody who knows about the often bumpy road of progress in AI and NLP, this must have been an amazing moment. Sure, it's a highly specialized machine, there's nothing universally intelligent about it, and "solving" Jeopardy might not mean much in the bigger picture (and I use the word "solving" in a very loose sense here -- but no matter what the outcome is, Watson has at least shown that he's able to play along with the best humans. Note the similarity to the first Deep Blue/Kasparov match.)

With all that said, this system is still an impressive (and highly visible) culmination of a long, long process of small advances made in the field in the past decades, and I think I should make the following known publicly now, right here, before it is too late:

I for one welcome our new robot overlords.

EDIT [29.03.2011] The original post date proves it, Jennings totally stole that line from me...