Computer program to take on world's best in Texas Hold 'em

[ahhming]

Carnegie Mellon University researchers are pitting a computer program against some of the world's best professional poker players.

Over the course of two weeks, the CMU computer program, Claudico, will play 20,000 hands of Heads-Up No-limit Texas Hold’em with each of the four poker pros. The pros — Doug Polk, Dong Kim, Bjorn Li and Jason Les — will receive appearance fees derived from a prize purse of $100,000 donated by Microsoft Research and by Rivers Casino. The Carnegie Mellon scientists will compete for something more precious.
 

“Poker is now a benchmark for artificial intelligence research, just as chess once was,” said Tuomas Sandholm, a professor of computer science at Carnegie Mellon who has led development of Claudico. “It’s a game of exceeding complexity that requires a machine to make decisions based on incomplete and often misleading information, thanks to bluffing, slow play and other decoys. And to win, the machine has to out-smart its human opponents.

 

"I think in the early stages, the computer might have an advantage, but at a certain point the players can figure out what's going on and adjust," Les said. "But it's a strategy game. You bring a strategy to the table and the computer's strategy may be better than mine."

 

"I hope we can stand up for humanity and take this computer down," he said with a laugh. "I know computers will eventually be able to beat humans. But I hope we can make them go a few more rounds after this before they do, like Kasparov did."
The computer will play 80,000 hands of Heads-Up No-Limit Texas Hold 'em in the two-week tournament.

Source:
http://www.cmu.edu/news/stories/archives/2015/april/computer-faces-poker-pros.html

http://phys.org/news/2015-04-world-texas-em.html

[Paralectic]

Interesting! cool

[Wind]

Are A.I is not complex enough?

[Sauron]

What about the random element? Isn't chess a better ground for experimenting an A.I.?

[Toddwjp]

What about the random element? Isn't chess a better ground for experimenting an A.I.?

 

because chess has already been solved by ai. poker introduces a complex system that will help develop AI in areas where it cant have all the information

[Sauron]

because chess has already been solved by ai. poker introduces a complex system that will help develop AI in areas where it cant have all the information

 

let it play starcraft then... in poker what the player does is very limited, and since an A.I. can't mimic facial expressions without it being obvious (or read any) even the bluffing aspect is reduced. On top of that, the game simply does not completely depend on you, and the best player in the world could loose to the worst one multiple times if the other guy is getting straight royal flushes.

[Trik'Stari]

An interesting thought, but couldn't the computer just count cards based on what it has in its own hand?

 

let it play starcraft then... in poker what the player does is very limited, and since an A.I. can't mimic facial expressions without it being obvious (or read any) even the bluffing aspect is reduced. On top of that, the game simply does not completely depend on you, and the best player in the world could loose to the worst one multiple times if the other guy is getting straight royal flushes.

 

Let it play DayZ and see how long before it gives up.

[Toddwjp]

let it play starcraft then... in poker what the player does is very limited, and since an A.I. can't mimic facial expressions without it being obvious (or read any) even the bluffing aspect is reduced. On top of that, the game simply does not completely depend on you, and the best player in the world could loose to the worst one multiple times if the other guy is getting straight royal flushes.

 

yes but that's why poker is a very good game for testing ai. because its not one hand or 10 hands. its thousands of hands against the same players. eg after 500 hands the hope is that the ai will be deciding how to play based on what it knows the other players have done in the past. it also will be observing the interaction between the players in hands they win. etc

[Sauron]

An interesting thought, but couldn't the computer just count cards based on what it has in its own hand?

 

 

Let it play DayZ and see how long before it gives up.

 

well bare in mind it would basically have aimbot and use exactly the amount of ammo it needs to without wasting any, it would also dodge anything that isn't a bullet and spot enemies as son as they come in vision range. So maybe it wouldn't be that bad at it ^^