Page 8 of 8
Re: Rerun of the Tromp-Taylor bet
Posted: Wed Jan 25, 2012 10:09 am
by Kirby
Mike Novack wrote:Kirby wrote:Though I am human, it is interesting to ponder how my brain works. Am I really capable of any cleverness from within, or is it all a result of my brain having been programmed?

And can depend on what you call "being programmed".
I suggest you look at descriptions of the type of AI that goes by the name of "neural net". These can learn (be trained) to perform a task but whether you would call this "being programmed" is another matter.
Although the hardware is very different, there are reasons to consider a "neural net" not entirely unlike our brains. At some level our "thinking" and what we have learned to do must be the result of connections between neurons reinforced or impeded. Interestingly, if you "damage" a nueral net the syptoms are not entirely unlike a stroke, especially relearning following a stroke (the neural net hasn't been physically damaged, no cells dead, just connections disrupted, so it can relearn).
Hehe. Actually, I'm familiar with neural nets a bit already. "Intelligent Systems" (basically A.I.) was the focus of my master's degree, and it is in fact the study of AI that has made me more curious about how my own mind works. The human brain seems quite complex, but there is a lot that is still a mystery. Is "artificial intelligence" always different than the "true intelligence" that humans have?
Re: Rerun of the Tromp-Taylor bet
Posted: Wed Jan 25, 2012 2:21 pm
by Suji
Mivo wrote:Suji wrote:In 20 years, who knows who'll win. Humanity will eventually lose to the top programs.
Clever programmers (human) will beat top players (human).

I see what you did there, but nonetheless I agree.
Mef wrote:Suji wrote: As of right now, chess engines are probably generally stronger than the best humans.
In 2009, a smartphone played at the level of a top grandmaster (~2900). Apparently they estimate best engines running on reasonable hardware are about 300-400 points stronger than the best humans. I think it's safe to say without a major breakthrough for us, the computers have claimed the peak of the mountain of chess. It will be interesting to see how long go holds out.
I'd like to see a match between Magnus Carlsen and Houdini 2 before I conclude that humans are weaker. Unfortunately, it's probably never going to happen. Carlsen would probably be the underdog, though.
Go will probably hold out another 20 to 30 years, maximum.
Re: Rerun of the Tromp-Taylor bet
Posted: Wed Jan 25, 2012 3:58 pm
by Mike Novack
Kirby wrote:..... curious about how my own mind works. The human brain seems quite complex, but there is a lot that is still a mystery. Is "artificial intelligence" always different than the "true intelligence" that humans have?......
I suppose it depends on the level at which you choose to look. A "program" has a certain independence from the "hardware" on which it runs. But at some level our "true intelligence" is just the result of selectively moving ions about triggering a resonse in some neighboring neuron(s).
And why just humans? Do you suppose we yet have a supercomputer controlled robot better locating and catching small prey than a cat? (and simultaneously evaluating possible threats to itself from larger predators). Natural intelligence (as opposed to artificial) isn't a humans preserve. The problems are quite complex (Is that other predator fit enough I need to worry? Is it interested in me or not? Is it perhaps faking indifference? Should I fake indifference to throw prey off guard? Or on the other side, should I hide that I know I am being stalked and am prepared to escape when the time comes?). )
Re: Rerun of the Tromp-Taylor bet
Posted: Fri Jan 27, 2012 11:59 am
by pookpooi
A better hardware is required to make the program play faster without drop in performance. Imagine Zen uses no more than 0.1 second per turn, it will put the pressure to human.
Re: Rerun of the Tromp-Taylor bet
Posted: Sun Jan 29, 2012 6:13 pm
by badukJr
It looks like Zen has been pulled off of KGS. What happened?
Re: Rerun of the Tromp-Taylor bet
Posted: Sun Jan 29, 2012 8:37 pm
by Joaz Banbeck
It heard that Magicwand was gunning for it.
Re: Rerun of the Tromp-Taylor bet
Posted: Mon Jan 30, 2012 12:43 am
by RBerenguel
Mef wrote:In 2009, a smartphone played at the level of a top grandmaster (~2900).
I'm writing a post side related to Go, and wanted to talk a little about Go and chess AIs in passing. Do you have any sources for this? I've Googled a little but found nothing, but my knowledge of online chess is close to nil (I gave up playing before having Internet at home)
Re: Rerun of the Tromp-Taylor bet
Posted: Mon Jan 30, 2012 1:22 pm
by Suji
RBerenguel wrote:Mef wrote:In 2009, a smartphone played at the level of a top grandmaster (~2900).
I'm writing a post side related to Go, and wanted to talk a little about Go and chess AIs in passing. Do you have any sources for this? I've Googled a little but found nothing, but my knowledge of online chess is close to nil (I gave up playing before having Internet at home)
http://en.wikipedia.org/wiki/Computer_chesshttp://en.wikipedia.org/wiki/Computer_GoYou can start with these. Also google the chess programming wiki.