Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

AI Beats Four Top Poker Players

bbc.co.uk

Post reply on HN
Print view View original

61–70 of 234 posts

Re: AI Beats Four Top Poker Players

#61

Post by arnarbi »

brilliantcode wrote: post #5

it won't be long before we hearing more headlines like: "AI beats top 10 hedge fund managers" to "AI run hedge fund blows up due to black swan event" regardless it's an incredible feat. It really casts questions into what our edge as humans are which is slowly disappearing and we didn't even need to put a brain in a jar and hook it up to a computer....it's deep learning reinforced algorithms that is appearing to outl…

> It's almost like we've uncovered ways to automate our intelligence very much like we've been automating human and animal labor in the past couple centuries.

Actually, a lot of mental work isn't based on intelligence but rather training in pattern recognition, thinking faster, translation of information between different coding systems, and developing instincts about the behavior if complex systems. This isn't very different from manual labor when you think about it.

Re: AI Beats Four Top Poker Players

#62

Post by JumpCrisscross »

philipov wrote: post #59
sulam wrote: post #55

Earlier quoted context omitted.

I was under the impression that you need some capital, but not so much that you move the market with your trades. True?

eh, if you've got the money to be worrying about moving the market, you're using brokers' advanced execution algos and darkpools.

> if you've got the money to be worrying about moving the market, you're using advanced execution strategies and darkpools

These aren't silver bullets. Both cost more, in fees, and have a habit of leaking order information across informal channels. They're also, generally, slower--there are, on average, fewer buyers and sellers for a given security in a dark pool than in the open market.

Re: AI Beats Four Top Poker Players

#63

Post by jedberg »

sanswork wrote: post #40
jedberg wrote: post #31

EDIT: Read below, I am wrong. I clearly didn't know what I was talking about. This is a great achievement in AI, don't get me wrong, but the headline should read, "AI beats the best four poker players we could find who were willing to play for a mere $200K". All the actual best players play for millions and have a reputation to uphold. They would never agree to do this. They four guys they got are pretty good, and co…

>All the actual best players play for millions and have a reputation to uphold. All these players are high stakes players. I think you're underestimating the fun factor. As for reputation. For a poker player having a reputation as being beatable is a profitable thing to have. >but they aren't the best of the best. Who do you think is? Like how many people do you think rank above this group at HUNL?

Anybody here? http://pokerdb.thehendonmob.com/rankings/

Or any of these people: https://en.wikipedia.org/wiki/List_of_World_Series_of_Poker_...

Re: AI Beats Four Top Poker Players

#65

Post by beambot »

reso wrote: post #15
feral wrote: post #7

Earlier quoted context omitted.

> "AI run hedge fund blows up due to black swan event" Knight Capital, 2012: https://www.bloomberg.com/news/articles/2012-08-02/knight-sh...

I believe the poster meant a machine-learning system blowing up (catastrophic trading loss) due to a black swan event that the system wasn't trained to deal with. Knight's blow-up was a process/sysops failure, not due to their strategies going wonky. [As a side note I believe their core strategies were human-analyst designed, not ML based, but I could be wrong.]

Wouldn't process & sysops failures count as blackswan events if the AI or ML algorithms weren't trained to deal with those types of hiccups?

Re: AI Beats Four Top Poker Players

#68

Post by sanswork »

jedberg wrote: post #63
sanswork wrote: post #40

Earlier quoted context omitted.

>All the actual best players play for millions and have a reputation to uphold. All these players are high stakes players. I think you're underestimating the fun factor. As for reputation. For a poker player having a reputation as being beatable is a profitable thing to have. >but they aren't the best of the best. Who do you think is? Like how many people do you think rank above this group at HUNL?

Anybody here? http://pokerdb.thehendonmob.com/rankings/ Or any of these people: https://en.wikipedia.org/wiki/List_of_World_Series_of_Poker_...

HUNL is a different game to MTT which makes up almost all of the ranking values for hendonmob and the WSOP ME winners specialities.

Re: AI Beats Four Top Poker Players

#69

Post by Fuzzwah »

jedberg wrote: post #63
sanswork wrote: post #40

Earlier quoted context omitted.

>All the actual best players play for millions and have a reputation to uphold. All these players are high stakes players. I think you're underestimating the fun factor. As for reputation. For a poker player having a reputation as being beatable is a profitable thing to have. >but they aren't the best of the best. Who do you think is? Like how many people do you think rank above this group at HUNL?

Anybody here? http://pokerdb.thehendonmob.com/rankings/ Or any of these people: https://en.wikipedia.org/wiki/List_of_World_Series_of_Poker_...

Tournament earnings are utterly unimportant in the world of Heads Up No Limit cash game poker.

Re: AI Beats Four Top Poker Players

#70

Post by bcassedy »

bko wrote: post #42

Does anyone know the technical details of how to train and AI bot for a game like poker? I imagine it's just reinforcement learning where the inputs are the actions of the individual players (hold/fold/raise, timing etc) and the statistical probabilities in terms of expected cards. Train a neural net to predict probability of the opponent's hands and act accordingly. Is it just that the professionals all act similarl…

I'm sure the actual process differs from research group to research group. Here's how one commercially available AI training system claims to work.

http://www.pokersnowie.com/about/technology-training.html

And they go into some of the problems that stem from how they've abstracted the game here:

http://www.pokersnowie.com/about/weaknesses.html

Post reply on HN