THIS!

And no one will ever be able to get anywhere (online). If you can write bots to win 2BB/100 at limits as low as nl50/ nl100,humans can't move up levels.

Mood has nothing to do with poker. It is related to the humans playing it. Also, in poker there isn't an infinite number of variables. But let's create a game in which there are an infinite number of possible decisions. It's a 2 player game and each player starts with 1 money. Each player takes it in turn to take any number of money (from 0 to 1 inclusive) from the other player. Whoever gets 2 money in total wins. On the first player's turn, he has an infinite number of possible moves. He could take 0 money, 0.1 money, 0.01 money, 0.001 money etc. But it's obvious that he should take 1 money because he will win the game instantly. So even though there may be an infinite number of possible decisions, it's not necesarily impossible to solve (or even difficult to solve, as shown by my example). It would be possible to solve the chess game you proposed.

I don't know enough about how the human brain works to comment on this. But anyway, solving poker would be ideal for an algorithm based thinking machine. It's just all mathematics. The human brain is great at making new inventions and being creative. No algorithm I know of can do that (or anything similar). If it tried to exploit opponents it would open weaknessees in its own strategy.

A lot of information is needed to be good at exploiting opponents. An AI would have to know how fast humans learn, how they tend to think, and a billion other things and how they all relate to eachother. It would be very difficult to create such an AI.

A computer will never have the edge barring the abilty to read emotions.
This also non withstanding that when the f%$k was poker ever a logical game.
I mean do you honestly think you can program a computer to switch gears from lag to knit based on what it percieves about the idiot its playing?
And still come away with a greater edge than you or i?
I mean i see what your saying ie chess programs ect..but when you also wiegh in the amount of random luck in poker vs chess ...apples and oranges my friend

The jury respectfully disagrees

the naivety in this thread deliciously lol-tastic...

anyone with half a clue about game theory, poker, and computers knows that it is a foregone conclusion that humans are sure to end up 2nd best against poker playing machines.

pray tell, non-believers: what kind of fairy dust is secreted from your pineal gland that gives you your magic poker insight that computer could never have?

we don't race horses against automobiles anymore either. get over it, humans... your brain isn't that special.

even *if* (and that is a big IF), this were true... what is to stop a computer from learning to read emotions?


please tell me it has something to do with a lack of a "soul" so I can begin the L M A O

I could care less what you have a degree in, Im also studying IT, mainly networking though, and ive met many people with high degrees that fail to understand simple things, for me its not your title but your skills that matters

And its total BS to say that emotions is "manifested" in data. If you for example look at the secario of a chess engine, which is more than anything a math problem. Garry Kasparov played vs. this with 2 loss, 1 win and 3 draws. 3draws and 1 win against this highly complex computer http://en.wikipedia.org/wiki/Deep_Blue_(chess_computer), was managed by kasparov because the machine will never play a perfect game of chess. It will look at huge amount of data of lines of moves, but the way its programmed its flawed because it favors a materialistic playing style more than anything else.

For example a player1 could have a "edge" vs player2 because of the given position, but the computer may state the game as equal or slightly in favor of player2. This is true in many ways, not only by means of a old engine like Deepblue which is from 1996, but also with the newest engines like Fritz. You may want to take a look at this game if you dont alleready know it. A game between Robert James "Bobby" Fischer and Robery Eugene Byrne, where not even Fritz did not see the winning line after Bobbys 18. Nxg2, and presumearbly bobby saw this four moves earlier when he sacrificed his knight on f2. This is one example where even modern and complex engines will not have the edge because of the way its coded.

Game: http://www.youtube.com/watch?v=brYQ5fhTS6o


My arguement is that a games which is obviously more based on math, chess, compared to poker, even engines can be wrong when they got huge amount of data avalible, in form of databases of games and huge amount of calculation / sec in terms of position and lines.



Im sorry for the long post, and inbetween bad english, I dont care to take time and review it and the comparation to chess is somewhat abstract, but I think you will understand my point - that yes in raw data computers are faar superior, but in more complex situations which is hard to understand for computers they still need faar more and better programming to understand.

Hope it all makes sense
Regards, Chilly

I voted no because of all the variables in poker, and they are all dependent on each other in making decisions.

By the above statement, I just don't see how we can make a computer strong enough to evaluate all variables. Just like I don't see how we can make a computer that will simulate a laboratory size experiment using using ab-initio methods with like billions of basis sets.

Yes computer power is increasing as does technology, but is it limitless?

All people arguing against are just ignoring that AI is already better than the best humans at heads-up limit hold 'em, amirite

I last longer than most guys.

lol PA being a good player

Voted yes. HU the AI wouldn't stand a chance, but if it would go shameless and take on the nosebleed degens with a 20BB stack in a 6max PLO game... need I say more?

Mark Vos already proved this, right?

Earth is flat because I just see it !
It can't be round because I don't see how it can be round !

This thread is amazing testimony of how clueless people are...

What many people are ignoring are there are two different notions of "solved", and people are using them here differently. There is "solved" in that you are playing an unexploitable strategy. An example of this is a computer that plays rock paper scissors by playing a random throw every turn. This strategy is unexploitable, and is break even against someone playing the exact same strategy.

Poker is, in fact, also solvable, which means there is a strategy that is completely unexploitable. But making a computer that does this really isn't that interesting, and it's for the same reason that the rock paper scissors bot isn't interesting either. If you were to pit that machine against a drooler who only played paper every throw, that machine would still break even, which is completely unimpressive.

So the interesting question is, can we (or future machines) create a bot that not only solves poker (impressive, but only from a number-crunching pov), but uses actual strategy and adaptation (and not just math) to exploit a given player? And the answer is yes. But it will take more than just improving computing power. That said, it's probably not all that far off given that technology improves upon itself exponentially.

Think it should be breakeven against any strategy

^^ Yeah sorry, obviously if you're playing an unexploitable strategy, it's unexploitable against any other strategy. So ya, the rock paper scissors unexploitable strategy is breakeven against any strategy.

Good points RML604, the analogy you gave is good, and a lot more relevant than comparing chess as it is also an imperfect information game.

thats like saying,

in the future, we are ruled by bots.

no way to answer questions like that.

how about, in 5 years, AI will beat a nosebleed pro headsup in 500k hands. something thats empirical...

While this is the case with RPS it isn't the case with holdem. It's very difficult to break even against optimal strategies in simplified holdem games (one street games for example with fixed several bet sizes) let alone full scale poker.

There won't be such day that we say "aha, NOW AI is better than humans". People who will be in possession of such bots may well not want for them to be public and nobody will be interested in playing 500k hands match vs AI anyway.

how do u think these research is conducted? it takes money and resources, some of the top AI research are done with government funding. not some nerd in his basement trying to make a bot that makes millions. you really think any significant breakthroughs won't be made public?

wat

I have no idea. What I know is that there are private teams of programmers working on it various poker AI's/analyzing tools. I have no idea if "breakthrough" will be made by guys from Alberta University or other public research center or by such private team. My money would be on private teams are there are usually the most motivated ones.
So I have no idea, but I would say that there is 99% chance that once NLHU bot capable of owning jungleman is made it won't be public for quite some time...

+1

its either impossible or possible as we speak...if it is not possible, it wont be possible in the future either.