Saturday, 28 July 2007
Smarter and smarter
In May 2009, Japan will adopt a jury-style system to settle criminal cases. Most Japanese, unsure about prospects for human error, say they dread the change. So they should, warns satirical United States commenator Stephen Colbert. He reckons this “jury of your peers” malarky is dangerously old-fashioned for such a hi-tech country.
“Keep in mind that through technological advances, robots are your peers.
“Robots have no mercy.” Creations of artificial intelligence do not yet sit in judgement on us, except perhaps for the tincture of of reproach in the beeps when you leave a fridge door open, or try to exit a car before it’s good and ready.
When it comes to combat, at least in its benign forms, the tide has surely turned. Canadian boffins have recorded what Science magazine considers a milestone in the development of artificial intelligence.
It goes by the name of Chinook.
Chinook is not programmed to shut down global defence mechanisms or to set steely Arnold Schwarzenegger lookalikes tromping their way to world domination.
At least that’s what the Canadians say.
Chinook does, however, play draughts better’n you.
It’s reportedly unbeatable, having figured out all 500 billion billion combinations. A perfect opponent would achieve a draw, but no pathetic human would defeat it.
We’ve long since lost the reliable upper hand in many such games.
It’s been a full decade since IBM supercomputer programme Deep Blue beat world chess champion Gary Kasparov.
Kasparov went down much like the legendary steel-driving man John Henry lost to a steam hammer.
Here is the full story.
Comparing the justice system to machines playing games is head-scratchingly pointless to me.
Just an FYI, Silicon (no “e”) is the element on which computer chips are based. Silicone (with “e”) is a synthetic rubbery substance often used for surgical implants.
Brad Hoehne
>>Kasparov went down much like the legendary steel-driving man John Henry lost to a steam hammer.>>
Whoever wrote that didn’t know much about chess. Kasparov could easily have won that match if he hadn’t self-destructed, and it was in fact IBM who quietly begged off playing a longer match. There was nothing inevitable about a human defeat in 1997.
Computing an exhaustive solution to a game (or to a position in a game such as a chess problem) is a classic task in my field of computational complexity theory.
For a view of how some professionals in my field reacted to the news, including a comment by me, see this entry in the Computational Complexity Weblog—which makes good reading for aspiring computer scientists, though with a bland layout compared to Susan’s :-).