Will computers dominate poker as they did with chess?

[Check2TheLady]

I accept your point, and it is a good one, but I would add that is not often a good idea to bluff completely randomly in poker. But neither is it a good idea to repeatedly bluff only when a scare card hits or only when your opponent's actions imply that they are weak. At the same time, a bot would have to bluff on occasion (and pick good spots to do so) or the pro would have an edge.

As I said before, a top player would be continually putting their opponent to the test, and a bot, by definition, is programmed to act in a certain way.

I would concede that these issues probably don't make it impossible for AI to be invented that could consisently beat a top player sometime in the future, but I do think that we are not even close at present. Every strategy has a counter-strategy, but the professional players have so many weapons in their strategic armouries and are so good at mixing up their play that any bot would have to be very sophisticated indeed to stand a chance against them.

Having said that, in heads-up play, going all in every hand would probably result in a fair proportion of wins for the bot. It depends really on whether we are talking about a bot designed to play one opponent or a full table of professionals? And if it is playing just one player, was it designed to beat that particular opponent's dominant style, and would the player have knowledge of either this or the fact that they are playing against a computer?

Sorry if this post appears a bit rambling, I am thinking as I type, but it is an interesting subject... [img]/images/graemlins/smile.gif[/img]

[jogsxyz]

[ QUOTE ]

Whereas I stand by my prediction that computers will never be able to beat a table of good human players in no-limit poker.


[/ QUOTE ]

With so much luck in poker, computers can beat humans now on a lucky day.
A fairer question is will computers be better players than humans?

Yes, because of the computer's record keeping abilities.
The computer can catalog all past play by its human opponents.
Computers will be programmed to play perfect game theory. The best EV action is often to bet the good hands and some of the bad hands. The computer will be able to randomize in an unpredictable fashion. Humans fall into patterns. Computers will be able to detect those patterns. The computer will be able to exploit the human and revert to optimal strategy when necessary.
The computer can do everything the humans can do and still do more.

[mr_hanky]

I think number of hands played is a crucial factor.

If the computer was tested by playing only 1000 hands against a line up of 1000 of the best players to make a total of 1,000,000 hands.- I think it would lose, because the human would be able to learn quicker over the small sample size.
If the computer started playing more that a few thousand hands against a single human it would be theoretically easier for the computer to make accurate predictions on the human. Then it may start to make very good decisions based on the complete recollection of all those hands.

[Blue Lagoon]

[ QUOTE ]
I think that the main reason that computers will not ever be able to beat humans at poker consistently is nothing to do with poker's luck factor or an ability/inability to learn. It is simply that they are too predictable. If a bot always makes the most +ev move in any given situation, it should not be too hard for the best human players to work out quite accurately what they are holding. Poker is a game of incomplete information, yes, but top players are extremely good at controlling the information that they give their opponents. It is also a game that involves a lot of deception and I do not think that a computer could possibly get the better of a player that properly puts them to the test.

[/ QUOTE ]

I don't think so. Look at "Optimal bluffing frequency" in Theory of Poker. You can tell your opponent that you will bluff 1 time out of 6 (if the ratio bet / pot size deserves it) and your opponent cannot gain an edge, by either folding or calling.
So it will not be a problem for a bot.

[jukofyork]

[ QUOTE ]
[ QUOTE ]
I think it's inevitable that they eventually will be able to beat any human at all forms of poker. In fact I'm fairly sure that eventually they will be able to beat humans at all forms of board games, card games, and even computer strategy games (which are much harder for AI due to their complexity).

[/ QUOTE ]
Whereas I stand by my prediction that computers will never be able to beat a table of good human players in no-limit poker.

Part of this is definitional. If the computer got sensory input from the humans, say by hooking each player up to a lie detector or using more advanced physiological sensors, the situation would be different. But I would argue that's no longer poker.

[/ QUOTE ]
I was only considering the abstract version of poker played out in a digital environment (much as today's online poker), but I'm not convinced adding sensory inputs will make it much harder for computers in the future and it could even make it easier. Machine Vision systems are only in their infancy and we've only fairy recently had hardware good enough to deal with the shear amount of data generated. It may turn out that future vision systems can watch all players at the table simultaneously at ridiculous frame-rates and be able to do things with the data we humans could only dream of. That's even ignoring the fact that computers may be able to add extra senses to their arsenal such as detecting tiny infra-red, micro-electric and/or chemical changes in the air (whether you would allow them to use these "extra senses" is another matter though).

[ QUOTE ]
Also, some day someone might learn how to grow a human brain in a jar, and train it to play poker. I'm talking about something resembling today's computers, but faster and with cleverer programs, working with inputs of the cards and bets only.

[/ QUOTE ]
If computers keep on advancing it may be possible to completely model a human brain using the very same Von Neumann architecture as today's computers use, but I doubt very much this would be necessary and by the time we ever have that level of understanding we will most likely be able to design vastly better machine learning algorithms than were designed by evolution alone.

I also don't think for one second it will be a machine "programmed" in the current sense of the word, and I'm pretty sure that even if every poker player worked together they couldn't "program" a gigantic expert system to play world class poker.

[ QUOTE ]
Playing a table is different from playing a hand heads-up. If all the other players conspire against you, you can't win however good you are. I think game theoretic play based on heads-up principles will naturally push all the other players to play to your disadvantage.

[/ QUOTE ]
I don't think the fact that other's can conspire against you is any different for a computer and this could just as easily be faced by a human. The fact that it complicates finding a game theoretic solution isn't a problem if presented with sufficient computing power. Even with the current state of algorithmic knowledge an algorithm could be created to find (near) "optimal" solutions to the n-player game by using some kind of evolutionary approach and/or fictitious play. If quantum computers ever arise then this may be possible much sooner than simply waiting for the current architecture to advance.

Juk [img]/images/graemlins/smile.gif[/img]

[jogsxyz]

The methodology for programming the computer to play chess and poker is completely different. After the chess opening chess program usually use brute force. Just try every move and see which one worked best. In poker there's all those known distributions. Preflop the computer just follows an opening chart. Postflop the computer assigns a distributional range for opponent's actions. This range will include the possibility of bluffs and semi-bluffs. Then run one million iterations. These calculations could be made in a second or two. The computer can always move instantaneously. Humans can't think that fast. Humans may have tempo tells. Computers don't. This ought to be a huge advantage for the computer.

[JocK]

[ QUOTE ]
what proof do you have of computers' learning abilities?

[/ QUOTE ]The strongest Backgammon bots are neural nets that started off by playing many games against itself using no more than some dummy strategy, making silly mistakes, whilst learning from these (i.e. adopting its strategy towards those that would have yielded better past results). No human-strategies or other game-theoretical information gets loaded into these bots. Yet, after a few hundred thousand games, these bots can not be beaten by any human except for a few world-class professionals...

[JocK]

[ QUOTE ]
Playing a table is different from playing a hand heads-up. If all the other players conspire against you, you can't win however good you are. I think game theoretic play based on heads-up principles will naturally push all the other players to play to your disadvantage. I don't think anyone has any idea how to build or program a computer to interact with humans. Maybe someone will do it someday, but it will take a fundamental breakthrough.

There's a lot more to poker than odds and strategy.

[/ QUOTE ]
Aaron, whilst you have strong doubts whether a computer could ever survive playing against a table of top-rated human, would you accept that in a number of years a top-rated human no longer can survive against a table of (independent) bots?

[jogsxyz]

Bill Chen with a team of crack programmers can probably be capable of beating all the best players now. The computer can unemotionally laydown strong losing hands much easier than any human player. The computer would never make a technical error. Have access to all the jam or fold charts. Make adjustments to the charts for each players' tendencies. Will there be the monetary incentive for anyone to create this program?

[btmagnetw]

[ QUOTE ]
Bill Chen with a team of crack programmers can probably be capable of beating all the best players now. The computer can unemotionally laydown strong losing hands much easier than any human player. The computer would never make a technical error. Have access to all the jam or fold charts. Make adjustments to the charts for each players' tendencies. Will there be the monetary incentive for anyone to create this program?

[/ QUOTE ]how many jam or fold decisions do you run into in non-donkament poker? and again, a computer playing optimally is not necessarily dominanting.