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

171–180 of 234 posts

Re: AI Beats Four Top Poker Players

#171

Post by coltnz »

fierycatnet wrote: post #166
bcassedy wrote: post #75

Anyone that's interested in reading a more detailed account of the experiment can do so here: http://www.pokerlistings.com/libratus-poker-ai-smokes-humans... The above article spells out some of the details of the competition. The winrate (14.72bb/100) that the AI achieved over the 120k hand sample is almost certainly not due to luck. It is a huge winrate that most pros have to employ strong game selection techniques…

That's pretty impressive, the win rate and amount of hands played. It's certainly not a fluke. At the same time, these 4 pros, I've never heard of them. This AI and organizers should focus on HU specialists. Perhaps even invite players like Negreanu, Ivey, and more recent online pros who've made a fortune. Just don't let Hellmuth to play this machine, it would be a disaster.

These pros are considered to be top 20 HU online specialists, they would destroy Negreanu and Ivey over the long run at this particular variant (which they have played millions of hands at).

Re: AI Beats Four Top Poker Players

#172

Post by maverick_iceman »

brilliantcode wrote: post #87
deepnotderp wrote: post #83

Earlier quoted context omitted.

> "...academic rigor..." "...calculus.." Dude, I'm not even asking you for anything that difficult here, how would you like it if I walked in and said "well, I don't need to learn any programing, my time is better spent magically making it work".

I found calculus and academia generally very difficult. I'm happy to know that you don't suffer from this terrible affliction which makes studying difficult. You don't need to know programming if you understand your workers and they understand it enough to convey to you the time & resource constraints. I have a million dollars for this project. What are the features and functions that will maximize my returns? Hire a…

>Hire a couple of smart engineers who already holds the technical knowledge

Then exactly what value are you bringing to the table? Money? Sales skills? If that's the case then why focus on ML/AI technologies? Any company needs money or sales expertise.

Re: AI Beats Four Top Poker Players

#173

Post by nippples »

Can someone clarify why exactly is this being presented as more impressive than beating Chess masters or even Go masters?

People talk a lot about number of states in poker, but the hand and visible cards can be easily (to a statitician) reduced to a scalar "probability of having the best hand".

At least in terms of crunching the possibility tree this should be a far less computationally intensive challenge.

Re: AI Beats Four Top Poker Players

#174

Post by maverick_iceman »

YeGoblynQueenne wrote: post #97
brilliantcode wrote: post #87

Earlier quoted context omitted.

I found calculus and academia generally very difficult. I'm happy to know that you don't suffer from this terrible affliction which makes studying difficult. You don't need to know programming if you understand your workers and they understand it enough to convey to you the time & resource constraints. I have a million dollars for this project. What are the features and functions that will maximize my returns? Hire a…

I think people are just saying that if you don't undertstand the technology a little bit yourself you'll never know who to hire. Maybe somebody looks like they know what they're talking about, but they don't. Maybe somebody really knows what they're talking about but they can't put that across very well. Maybe someome is simply unscrupulous and wants to milk you dry and will dazzle you with charts and graphs that the…

Absolutely. I know just enough software engineering to bullshit my way in front of a nontechnical person. If any such person hires me to do serious dev work they'll be in a sea of trouble.

Re: AI Beats Four Top Poker Players

#175

Post by andrewprock »

RivieraKid wrote: post #113
andrewprock wrote: post #56

Earlier quoted context omitted.

I'm surprised that Andrew Ng made this claim. The strategy that was built for Libratus' predecessor did not do sophisticated modeling of the opponents, or use new algorithmic principles. Poker is solved using a very large game tree, just as with the other games. The structure of the tree is modified to support the notion of hidden state, but beyond that it is essentially the same as the other games. The structure for…

You have to model opponent's behavior though. For example, the probability opponent will fold when AI bets a certain amount.

This is sort of correct. In a Nash equilibrium, the other player is modeled as also playing a Nash equilibrium strategy. Nothing more.

Re: AI Beats Four Top Poker Players

#176

Post by maverick_iceman »

brilliantcode wrote: post #142
jzwinck wrote: post #133

Earlier quoted context omitted.

> Your top sales guy makes $9 for every $1 spent on him (+$8) where as your top engineer costs $3 for every $0 he generates (-$3) In this case you should immediately fire all your engineers, your CFO, and your COO. Because why would you want to employ a team of engineers who are costing you money but bringing you no value? You should also fire your cleaning staff, who have never closed a deal in their lives, plus the…

I'm glad you posted it because it shows how little people understand their engineer's financial importance. Granted, engineers are required to create products and maintain it. You need product to sell. But at the marginal level for every dollar your sales person makes your engineer is not. You might argue but the product is generating revenues but that's not what drives a sale. A sale is a function of value derived f…

If good engineers are dime a dozen then why do they command >200k salary? I don't think you understand how much skill and practice goes into making a good programmer. Your ignorance is truly astonishing.

Re: AI Beats Four Top Poker Players

#177

Post by maverick_iceman »

brilliantcode wrote: post #135
xapata wrote: post #120

Earlier quoted context omitted.

Funny, I know more people who are asking about machine learning than people who are experts. So who is in high supply?

Machine learning experts are in higher supply, your anecdotal experience offers little weight. High demand creates a high supply which drives wages to zero.

I think you are just trolling at this point.

Re: AI Beats Four Top Poker Players

#178

Post by roenxi »

nippples wrote: post #173

Can someone clarify why exactly is this being presented as more impressive than beating Chess masters or even Go masters? People talk a lot about number of states in poker, but the hand and visible cards can be easily (to a statitician) reduced to a scalar "probability of having the best hand". At least in terms of crunching the possibility tree this should be a far less computationally intensive challenge.

What struck me looking at the AlphaGo papers was how human its algorithms were - it was basically doing what a highly orthadox human player would do (pick some moves that look interesting, read as deeply as possible) and then scaling up with a completely inhuman thoroughness and accuracy.

At this stage, AI wins in competitive games simply will not impress me. From here it may simply be a tour of force showing the breadth of fields AI can dominate in.

After that enthusiastic agreement; poker wins are evidence for the laypeople of something I suspect most people in AI research or game theory already know - a human bluffing isn't a magic advantage in a fair game. The AI can outperform on fundamentals.

Re: AI Beats Four Top Poker Players

#179

Post by marklgr »

coltnz wrote: post #171
fierycatnet wrote: post #166

Earlier quoted context omitted.

That's pretty impressive, the win rate and amount of hands played. It's certainly not a fluke. At the same time, these 4 pros, I've never heard of them. This AI and organizers should focus on HU specialists. Perhaps even invite players like Negreanu, Ivey, and more recent online pros who've made a fortune. Just don't let Hellmuth to play this machine, it would be a disaster.

These pros are considered to be top 20 HU online specialists, they would destroy Negreanu and Ivey over the long run at this particular variant (which they have played millions of hands at).

Back when I played poker (up to ~2010), Negreanu was mostly a celebrity, as well as a pretty good blog writer. He was no match for the best pro players, but perhaps he has dramatically improved.

Ivey was pretty good at managing his image, too.

Re: AI Beats Four Top Poker Players

#180

Post by orless »

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…

> So the question is, how does an average joe hacker like me exploit and leverage this wonderful thing called deep learning? I'm not interested in reading PHD papers with advanced calculus.

You've got a lot of comments like "if you're not willing to read a paper with BASIC CALCULUS ... then I'm not sure machine learning is the right place for you" to this, but I second you on this. I do have a PhD in CS and not afraid of calculus in papers. I just don't think reading PhD papers is the best path to getting practical results in DL/ML. "Not right place for you" is the sign of the "exclusiveness" problem.

I'd recommend checking the following course by Jeremy Howard and Rachel Thomas:

http://course.fast.ai/

It is designed to make DL accessible and achieves its goal very well. I'm not sure if the average Joe Hacker will be able to write top-pocker-players-beating program after this course, but it definitely gives enough insight and practice to get started in DL and be able to design and code solutions in a large number of problem classess.

Post reply on HN