Humanity In (and On) Jeopardy
There’s a lot of hubbub right now about Watson conquering Jeopardy, and I wanted to step in to help put some of the fire out.
First of all, though, I find it odd that people don’t really understand that the interesting bit about this win isn’t related to the fact that the computer knows the answer to the question – this is no more impressive than discovering a set of paper-based Britannica Encyclopedias contains the approximate population of Malaysia.
The tricky bit for the engineers was in teaching the computer to interpret the format in which Jeopardy presents its questions, (answers, really,) so that it could dig through its on-board memory for the correct nugget. As Barry, over at Mr Blog’s recently mentioned, you can’t stuff any old search term into something like Wikipedia in the hopes of getting the right answer – even Google is just making its best guess when you start looking for “Mexican skull recipe”.
Watson’s creators needed to find a way to take that textual clue, then backtrack it to a reasonable answer – and they did.
So, now that a self contained search-bot has defeated us in the most critical of arenas, Jeopardy, what do we do? Give up? Start betting on competing trivia-computers? Search for John Connor?
No.
The time has come for humanity to get serious about next steps, and I don’t just mean as it relates to television game shows. Computers have defeated us in chess, Jeopardy, and Street Fighter II, but it isn’t that they’re somehow a malevolent force that’s putting us out to pasture, instead we need to consider how we can use personalized technologies to better augment ourselves.
Could Watson have won so handily if Ken Jennings had been allowed a laptop, with wifi access, on his podium? Probably, because fingers are meaty and slow – but could it have won if Ken Jennings had an interface implanted directly into his brain?
Suddenly it becomes a race related to reasoning, and not simply data correlation.
Are we going to cancel baseball once we discover that wheeled robots – with high-end sensor suites and mechanical reflexes – can out-run, out-bat, and out-catch, a human? Or are we going to slap some pistons into Derek Jeter and move the home-run fence back a bit?
Ken Jennings wrote an Op-Ed in today’s NY Daily News. I attribute it in part to sour grapes (Ken is not the nicest person, as Jeopardy! crew members can attest) but he has a couple of valids point, ones I wondered about as I watched.
http://www.nydailynews.com/opinions/2011/02/17/2011-02-17_ken_jennings_exclusive_oped_jeopardy_champ_says_computer_nemesis_watson_had_unfa.html
Very interesting article – I’ve heard similar about Jennings, but I feel like that all goes back to The George C Scott Problem.
(http://skinner.fm/2010/11/10/tut-teutul-the-george-c-scott-problem/)
Jennings makes some interesting points – although, frankly, I too suffer from “train-wreck categories” playing along at home.
I once had an extraordinary run of Jeopardy! luck while at a laundromat and pissed off a guy to no end, but that’s an old blog.
Should there be any comment lurkers lying about, I believe that excellent bit of bloggery can be found at:
http://bmj2k.com/2009/11/13/pathetic-paper-airplanes/