Humans Hold Their Own At Poker, For Now
from the one-last-redoubt dept
Score one for humanity. After four grueling sessions of human vs. computer poker, the human players managed to outlast and outwit the bot. It was by no means an easy task, as the computer performed strongly on day one. But, by the second day, the human players seemed to have a pretty good feel for the way the machine played the game, giving them the leg up. Assuming that computer, dubbed Polaris, represents the vanguard of AI poker, it would seem that it'll still going to be a long time before computers can compete at the top level, as they can do in chess. This particular match offered the computer optimal conditions, as it faced an opponent heads up in limit poker; in a no-limit game with a full table of opponents (conditions that would make it even harder to calculate proper strategy), its performance would have likely been significantly worse.
Thank you for reading this Techdirt post. With so many things competing for everyone’s attention these days, we really appreciate you giving us your time. We work hard every day to put quality content out there for our community.
Techdirt is one of the few remaining truly independent media outlets. We do not have a giant corporation behind us, and we rely heavily on our community to support us, in an age when advertisers are increasingly uninterested in sponsoring small, independent sites — especially a site like ours that is unwilling to pull punches in its reporting and analysis.
While other websites have resorted to paywalls, registration requirements, and increasingly annoying/intrusive advertising, we have always kept Techdirt open and available to anyone. But in order to continue doing so, we need your support. We offer a variety of ways for our readers to support us, from direct donations to special subscriptions and cool merchandise — and every little bit helps. Thank you.
–The Techdirt Team
Filed Under: ai, game algorithms, poker
Reader Comments
Subscribe: RSS
View by: Time | Thread
you play the man not the cards. How's a computer going to do that?
FIRST
[ link to this | view in chronology ]
SECOND!!!!
When you hear the hard drive clicking...it means there's no way he's betting on this hand...That or it is about to Crash.
Either way, Congrats to the humanoids!
[ link to this | view in chronology ]
Sure, it might use a 100% logical approach, but would it bluff, or attempt to?
How much would it risk in doing so?
Interesting concept..
[ link to this | view in chronology ]
Poker Computer
[ link to this | view in chronology ]
Poker, on the other hand, requires a significant amount of human factor. Bluffing, reading body language, playing mind games, etc., are all human traits, which are difficult for a computer to understand. There are most likely hundreds, if not thousands of subtle factors that a computer would have to consider, which makes for extremely complex calculations. A computer that can analyze all that data and adapt to each human player it is up against will be virtually impossible, especially if the human players can adapt to the computer faster than it can adapt to humans.
And finally, the factor of randomization makes it just that more complex, because unlike a chess or checkers game where every move is cumulative from the beginning of the game to the end, every hand in poker is like starting over fresh, which makes it very hard to calculate one's next move.
[ link to this | view in chronology ]
Re: waste of time
The goal is not to produce a poker playing robot. The goal is to develop artificial intelligence. Just like the computer program that will never loose at checkers. Why would you want to play it? Useless, but the goal was to solve the problem and learn something along the way.
[ link to this | view in chronology ]
What is very interesting about this though is it means perfect strategy does not win at poker. It would be interesting to see what would happen playing the computer against itself over extended periods of time. It would be able to play incredibly quickly as the human factor would be removed. With a large enough starting amount to eliminate short term variance, they should never be able to beat one another.
[ link to this | view in chronology ]
but games like poker the game is different. you see a portion of the gaame pieces, and your opponent sees another set. sometimes you have a community set. taking 3 minutes to place a bet can be bluffing, or strategy. either way there is no way to set a rule that if a players waits x seconds, this is what they are doing. now if you could track what the player did at the end of each hand, it might make things a bit easier, but only against that one player. i.e. now it's important to know how long decisions are made.
and ontop of that, there are small signs of how players react. subconsciencely, the way they put their chips, play with their chips, when/where they look...
it's just that poker has too many imputs/variables.
chess has 16 per side. checkers has what? 12 pieces a side. fairly easy to "know" the move sequience.
but with poker, hands of 5, using 5,7 or more cards per hand though... just a bit too much information right now.
[ link to this | view in chronology ]
missing
Luck is a very big part of the game as all of the play is determined by chance. No starting set positions, not the luck that your opponent makes a mistake, but that the game is designed for gambling, and luck is the backbone.
Two - Risk/Reward
Poker at its heart is about risk vs. reward. You can talk about bluffing and tells and stuff but that's not how successful players make their money. It is about pot odds and when bets make sense. If i'm likely to win a bet 20 percent of the time and it pays better than 5:1 I should make the bet. If I could do so in perpetutity I would basically be printing money. How you figure that out is the tough part and that's what the players are getting to know about their computer advesary.
[ link to this | view in chronology ]
Re: missing (luck)
[ link to this | view in chronology ]
The computer knew how to bluff guys
[ link to this | view in chronology ]
From one of the Players
Even more exciting than that however, is the potential for cooperative reactions from software and computer systems. Again, if you can predict what a human is planning, then you can also cooperate with those plans and provide up resources to help the person. This can lead to innovative and intelligent user interfaces, etc.
-Ali Eslami
indigoae@hotmail.com
[ link to this | view in chronology ]
[ link to this | view in chronology ]
Agreed--Resistance Is Futile
The software WILL become a superior player eventually, either through brutal statistical sleight-of-hand or by emulating the emotional responses of players to model them--and presenting its own "emotions" to baffle them. What's more, it might be taught to alter its "tells" over time to further confound its human opponents.
Sorry folks; we're still just really smart monkeys, and until we can artificially augment our own intelligence, our brain-children the computers will begin to meet and exceed us in ever-expanding fields.
[ link to this | view in chronology ]
AI...
[ link to this | view in chronology ]
That would be nice
[ link to this | view in chronology ]
It was interesting to read it
[ link to this | view in chronology ]
[ link to this | view in chronology ]
When you hear the hard drive clicking...it means there's no way he's betting on this hand...That or it is about to Crash.
[ link to this | view in chronology ]
What is very interesting about this though is it means perfect strategy does not win at poker. It would be interesting to see what would happen playing the computer against itself over extended periods of time. It would be able to play incredibly quickly as the human factor would be removed. With a large enough starting amount to eliminate short term variance, they should never be able to beat one another.
[ link to this | view in chronology ]