Poker AI - Page 5

David Steele · #41 12-10-2003, 03:15 PM

"BUT i don't beleive that it could ever ... "

This is the point you lose your credibility.

I believe that beating low limit games is not particularly
hard if it hasn't already been accomplished, beating any games is certainly possible too, people are not magical geniuses at the center of the inteligence universe.

D.

brianmarc · #42 12-10-2003, 03:22 PM

Yep-Your near unintelligible post certainly makes a good case for your argument about the lack of human intelligence in the universe

eastbay · #43 12-10-2003, 03:26 PM

[ QUOTE ]
I do agree that a program could be made to play agaisnt and beat low limit hold'em BUT i don't beleive that it could ever beat any skilled player or group of skilled players, especially short handed. The reason is this.

[/ QUOTE ]

Well you're wrong. It's been done. Heads-up against expert players.

Nottom · #44 12-10-2003, 05:20 PM

You could probably make a case for the fact that a BOT would be able to play heads-up better than a full game. Game theory based bluffing and randomization combined with the ability to calculate relative hand strength vs all possible holdings on the fly could make a well written comp opponent very nasty indeed.

jknupp · #45 12-10-2003, 05:52 PM

Why? Carengie Mellon has the number two computer science program in the country, behind MIT. It's in front of schools like Stanford and Berkley. What made you lose interest, the school, or the fact that I'm a student? All I was saying by that statement is that I know more about computer science than Joe Average. Why flame?

David Steele · #46 12-10-2003, 06:04 PM

What part didn't you understand and why a personal attack?

I just am particularly jolted by a claim that a computer will
NEVER do some task without any theoretical argument. Especially when in my own estimation the task is not that hard, and also when there are many references here to work where the task has been solved to some degree.

D.

John Feeney · #47 12-10-2003, 09:49 PM

Don't know if this has been mentioned in the thread, but it's well known that, by all indications, some very strong heads-up bots have already been made. They played for some time (maybe about a year?) on Paradise Poker a few years ago until, apparently, they were kicked off. AFAIK, they did quite well, and were competitive even against stronger players. I heard they were written by some folks in Switzerland who had previously written some strong backgammon programs. That said, "AFAIK" isn't very far, but that seemed to be the case from everything I heard at the time.

This has come up in past threads on this subject. A search might turn something up.

Dylan Wade · #48 12-12-2003, 05:43 AM

I've thought of a good idea for a braindead poker bot that is sure to win. Basically, find these seven card stud games online with no-ante. (though, there is a bring-in) You can give it the most brain-dead AI and it'll grind out some $$.

Al Schoonmaker · #49 12-18-2003, 09:24 PM

The current issue of The New Scientist, a British publication, has an article on this subject. It refers primarily to the work being done at U of Alberta, and quotes David Sklansky, Barry Tanenbaum, and myself.
Regards,
Al

Lunamondo · #50 12-19-2003, 03:52 AM

According to the Poki demo software (that I just downloaded and tested), they (bots) are not smart enough to be considered good players. Though I didn't got a chance to configure good solid bots with that demo, but I doubt it will drop the many mistakes and blunders I saw them making. They do think like humans, but trying to make them think so as smart as humans might not happen, even though I don't see there is any other way to finally get them as smart as engines are at chess where they don't have complete knowledge either but use their best intelligence to make the best moves. The AI poker softwares don't seem to be situationally all that much smarter than the more or less mere profiles.