Kramnik will lose game six and the match; for the human that is two losses and four draws. Kramnik’s main comment has been that the computer is hard to beat. With the exception of one major blunder he has played remarkably well and most human grandmasters would have lost almost every game.
Bah, any hunk of silicon (or carbon) can play Chess. Now Go, that is a game for a thinking being. Computers such at Go.
Moxy, try this one instead:
http://susanpolgar.blogspot.com/2006/12/game-6-fritz-1-0-kramnik-to-win-match.html
After Go, what next?
52-card pickup. Try *that*, mainframe!
Humanity has failed.
Humanity has failed.
Michael, gimme a break!
“Whether computers are smarter… not by a long shot.”
This is certainly true, although of course it depends on the definition of long shot. Computers however are getting a lot smarter , and humans, not so much. Although I think Tyler’s concept of siloed exponential growth in econ is not a model of reality, in this case it really is. I expect that in my or my children’s lifetime we will see the day, and it oughta change things quite a bit.
Hi, A Tykhyy. I’m a lot more optimistic, if that’s the right word. Even if people haven’t mastered it, I think computers will, when they effectively “learn” it, whether by using neural nets or genetic algorithms or what not. However! I’m completely ignorant about this great game. This has made me want to learn about it, though.
asg: The goalposts of what computers can do. I think we’re seeing that computers can play chess better now, given the record in recent matches. So for those who said a machine would never beat the world chess champion, that appears not to be the case. Now people may suggest that a computer will never beat Go champions.
I don’t think a computer beating the best at Go will necessarily be any more intelligent than one that beats Chess grandmasters today, but I think it will be done. There is a simple metric here: win or lose. Did the computer beat me at chess? Did the computer beat me at Go? There isn’t much to debate about better if you can go by 1s and 0s.
Debating about actual “intelligence,” whatever that means, is whole ‘nother story, of course.
Where did I read recently that it might be a lot like the airplane. People assumed that the route to flight would be by copying the flight of birds, but it turned out to be quite different. What we think of as intelligence today might similarly be turned on its ear when our machines take off and become “smarter” than us to all intents and purposes, although they might think differently. Even so, I’m sure we’ll want to keep redefining the meaning of intelligence to maintain our belief in ourselves as the pinnacle of creation.
Sebastian: Provided the computer does not cheat, even the declarer play is not great. It takes the best play in pure math terms, e.g. finesse vs. drop, but even good bridge programs have trouble with unblocking and hold-ups, to say nothing of endplays. As you suggest, the bidding is horrific.
Scott: Ironically, the best Go programs today are the ones that try to emulate human play, rather than algorithmically generating a strategy. They code in the behavior of human players, given certain board situations, and try to apply the rules. Of course, many Go positions are mathematically fundamentally different from chess positions (see the Wikipedia piece someone else linked). But if all we’re talking about is playing a game, then sure, great, computers are better than people are at chess now. What does that say about intelligence? Zero.
Comments on this entry are closed.