Open Side Menu Go to the Top
Register
In the future... AI > Best Players? In the future... AI > Best Players?
View Poll Results: In the future... AI > Best Players?
Yes
296 53.24%
No
260 46.76%

09-12-2010 , 12:59 AM
Quote:
Originally Posted by Shpank_oh
Computer processing ability will rival that of the human mind in ~2035, so yeah I think it's just kind of a matter of time tbh.
Doubt it
In the future... AI > Best Players? Quote
09-12-2010 , 01:10 AM
Quote:
Originally Posted by MetalSpork
Why does everyone think bots that clever would even want to play poker?
In the future... AI > Best Players? Quote
09-12-2010 , 02:52 AM
Quote:
Originally Posted by Hirle
Elisha Cuthbert, from "The Girl Next Door". Personally one of my favourite films :P

I think yes, one day AI will be able to beat those players.
In the future... AI > Best Players? Quote
09-12-2010 , 03:14 AM
It's funny how all the people who probably know nothing about AI or computer programming are all on the "IT WILL HAPPEN IN THE FUTURE ONE DAY" mantra. I think that's referencing technological advances that allow this? Definitely not the case.

I could write up a really good program, where the player is a 14/12/3AF, 5%3b, steals in perfect situations, 4b bluffs for balance etc. But the problem isn't with preflop, it's post. The situations that occur post are the reason so many players still BE or barely win at this game even though they have 2+2 and video training sites telling them exactly how to play the game. No bot can be written to handle the amount of factors that can occur post beyond bet / check / fold. The levels of thinking alone would be impossibly hard to account for and our bot would just get exploited b/c it's so clearly going to be on level 0 most of the time, with some random bluffs for "balance".

What if we don't have a big sample on a certain villain for our bot to make a calculated decision on? Like if this particular villain has a 50% raise c-bet stat over 5k hands but we only have 10 hands on him? How does the bot handle having TP in this spot? What about a good draw? The entire game is too variable dependent for any real calculated move to be "right".
In the future... AI > Best Players? Quote
09-12-2010 , 03:17 AM
in 50 yrs pc's will be alot smarter then humans could ever become.
In the future... AI > Best Players? Quote
09-12-2010 , 03:27 AM
I guess we will never find out.
In the future... AI > Best Players? Quote
09-12-2010 , 03:49 PM
Quote:
Originally Posted by Immortal-1337
Doubt you got a clue about programming, it`s still an alogorithm. If u got the algortihm you cou could know which cards you are dealt etc.

edit: I try to find a good article about it.
Don't really see how this is at all relevant to what he was saying.

Quote:
Originally Posted by Shpank_oh
Computer processing ability will rival that of the human mind in ~2035, so yeah I think it's just kind of a matter of time tbh.
Perhaps. However, I can't really see anyone being able to program something which can mimic a human brain.
In the future... AI > Best Players? Quote
09-12-2010 , 04:36 PM
Quote:
Originally Posted by zuk1
Perhaps. However, I can't really see anyone being able to program something which can mimic a human brain.
That statement had nothing to do with the discussion either. With enough funds and effort, we could easily program a bot able to beat Phil Ivey with today's technology. It has nothing to do with raw computer power. But it won't happen because the incentives aren't there, i.e. to program such a bot you need to be very knowledgeable yourself in poker, and why not then play directly instead? You avoid many potential legal issues.
In the future... AI > Best Players? Quote
09-12-2010 , 04:56 PM
I find it highly unlikely that it's at all possible to create a bot who's skills surpass the programmer's.
In the future... AI > Best Players? Quote
09-12-2010 , 05:02 PM
Quote:
Originally Posted by o22
I find it highly unlikely that it's at all possible to create a bot who's skills surpass the programmer's.
See it as a "superhud" which better process the vast amount of info available, can adapt instantly to its opponents, never tilt, and introduces randomness in its play.
In the future... AI > Best Players? Quote
09-12-2010 , 05:37 PM
Excluding the tilt issue, none of those things are anything a decent player with a HUD can't do.
In the future... AI > Best Players? Quote
09-12-2010 , 05:40 PM
Quote:
Originally Posted by o22
I find it highly unlikely that it's at all possible to create a bot who's skills surpass the programmer's.
"Robot Scientist Becomes First Machine To Discover New Scientific Knowledge"

http://www.sciencedaily.com/releases...0402143451.htm
In the future... AI > Best Players? Quote
09-12-2010 , 06:06 PM
Quote:
Originally Posted by o22
Excluding the tilt issue, none of those things are anything a decent player with a HUD can't do.
Even players like Ivey don't make the most +EV decisions all the time.
In the future... AI > Best Players? Quote
09-12-2010 , 06:07 PM
how can a bot spot a perfect bluff-squeeze-spot ? and then addaped right if they dont hit the board. or how can a bot complete donk any flop from the sb without beeing exploitable in longterm ?

nlhe has soo many differend moves and patterns and so on that i dont think a bot/a log would be able to exploit everyone.

like the 52o hand ivey vs lex. there is no way the bot could know that ivey was punishing greenstein and at the same time know that lex could be beware of that. so mathematilcy the bot would have to fold b/c obv calling a 5bet shove with KJ is -ev..
In the future... AI > Best Players? Quote
09-12-2010 , 06:13 PM
Quote:
Originally Posted by dontbescurrrred
how can a bot spot a perfect bluff-squeeze-spot ? and then addaped right if they dont hit the board. or how can a bot complete donk any flop from the sb without beeing exploitable in longterm ?
Those are all pretty easy to implement.
In the future... AI > Best Players? Quote
09-12-2010 , 06:22 PM
I'm not worried about it in the short term, but look how far computing has come in the last 10 years. It's a pretty safe assumption that at some point computers will be able to surpass the human mind in computing as well as perfectly emulate and understand human nature within the next 20 - 50 years. It's a lock.
In the future... AI > Best Players? Quote
09-12-2010 , 06:26 PM
There are supercomputers today which are way beyond unbeatable in chess and games like that. Poker is obviously a more situational and much more complicated game for a computer to adapt to, but seeing how fast the world has evolved the latest 20 years I wouldn't wager much money against this happening.
In the future... AI > Best Players? Quote
09-12-2010 , 06:34 PM
Durr= AI. didn't you read that article about the chip in his head?
In the future... AI > Best Players? Quote
09-12-2010 , 06:48 PM
I think too many people equate god tier poker bot to X years we will have super computer that will blow Ivey away. We have computers out now that could run billions of bots processing millions of a hands a second. The idea that a bot needs to "out pace human thinking" is illogical b/c humans have emotion and other variables that make our thinking so creatively genius while programmers, no matter how creative, could never code a bot to be so dynamic.

Now you could easily make a bot that could beat fish, but skilled players > bots any day.
In the future... AI > Best Players? Quote
09-12-2010 , 07:01 PM
Quote:
Originally Posted by xplore111
The idea that a bot needs to "out pace human thinking" is illogical b/c humans have emotion and other variables that make our thinking so creatively genius while programmers, no matter how creative, could never code a bot to be so dynamic.
And how does that apply to a game like Poker? Even if the human brain can handle a large number of variables and factors, it cannot compare to a computer, which has a virtually perfect memory and infinite processing power. And don't get me started on creativity in poker: if you apply reductionism, you'll find in the end it's only a matter of +EV or -EV decisions...

With a team of poker pros consultants and the best programmers available, it could be done in weeks.
In the future... AI > Best Players? Quote
09-12-2010 , 07:08 PM
Quote:
Originally Posted by whininguser
What level of NL games can the best bot beat these days?

I think bots can beat the top games only if they can mimic the thought process of top players, but I could easily be wrong because I'm no programmer and I don't know if it's possible to create a program that doesn't "think" like poker players but can outsmart them nonetheless.
they were some of the best regs at NL1K:
http://forumserver.twoplustwo.com/56...ipoker-766070/

I'd say that bot probably can beat NL2K these days...
In the future... AI > Best Players? Quote
09-12-2010 , 11:12 PM
The people saying no are greatly underestimating the power of computers in the future imo.
In the future... AI > Best Players? Quote
09-13-2010 , 03:26 AM
Quote:
Originally Posted by Acekob
There are supercomputers today which are way beyond unbeatable in chess and games like that. Poker is obviously a more situational and much more complicated game for a computer to adapt to, but seeing how fast the world has evolved the latest 20 years I wouldn't wager much money against this happening.
Pretty silly comparison I think, there's a huge difference between chess-playing computers and poker-playing computers. With chess-playing computers, people twenty years ago could have written the algorithms they're using today- the limiting factor was processing power. Computers play chess using brute force processing to evaluate zillions of situations at once and that makes them very powerful- but you can't do that with poker at all. The problem with poker isn't that we know how to solve it but can't, it's that nobody can even say what a solution would look like. I mean, we don't even know what the optimal opening size is in a HU NLHE match- some top pros will start off minraising, others will be 3xing, still others might mix it up, there's no consensus even for such a "simple" problem. You think you can just throw processing power at that until it fixes itself? I think if that were possible someone would've done it by now.
In the future... AI > Best Players? Quote
09-13-2010 , 03:39 AM
I would be surprised if this wasn't happening already. "Fuzzy logic" code deep-diving hand history dbs and stat dbs, throw in some networked bots and you're good to go. If I could program for s**t I'd be tempted.
In the future... AI > Best Players? Quote
09-13-2010 , 04:01 AM
A lot of you ppl flatter yourselves. I watched some expert of the TEEVEE say that it's not that computers will be smarter than ppl it's that they will be millions of times smarter. Some other expert said 'Hopefully, they'll want to keep us as pets.' And then there are the ppl saying that tech is nothing less than an evolutionary leap. Think about it for a minute: Many, many of the world's most brilliant ppl are working w/ virtually unlimited budgets to create machines that can do what ppl do and do it better. That's the reason that computer was built to beat the world chess champ.

And here's some random article:

http://www.foxnews.com/story/0,2933,296324,00.html

Any of you geeks liked 'Dune'? The Butlerian Jihad may be on the way.

The biggest lol to me is that the vote was tied when I voted.
In the future... AI > Best Players? Quote

      
m