На главнуюЛюди и блогиПохожее видеоЕще от: Doug Polk Poker

The Fight For Humanity Rages On!

Оценок: 1428 | Просмотров: 49606
Doug takes a trip to Pittsburgh and joins his close friends (and former students) Dong Kim, Jason Les, Daniel McAulay, and Jimmy Chou to talk about the Brains vs. AI challenge. They discuss their difficult experience playing heads up against a supercomputer as well as their findings of its strange playing behavior. Check out Doug's training website Upswing Poker: https://www.upswingpoker.com/ FOLLOW DOUG: Snapchat ► http://snapchat.com/add/WCGRider Twitter ► https://twitter.com/DougPolkPoker Instagram ► http://instagram.com/DougPolkPoker Facebook ► https://facebook.com/DougPolkPoker Twitch ► https://twitch.tv/DougPolkPoker #GrindNation
Категория: Люди и блоги

Html code for embedding videos on your blog
Текстовые комментарии (272)
willcarism (1 месяц назад)
DAN Polk... hmmm... still don't know what to think about this.
R. Veda (1 год назад)
Sounds like one of those 'statistical ties' you hear about.... ;-)
SuperDangerousMouse (1 год назад)
Not sure if it's worth it, but can you make another video if the poker community learned more about the AI and Libratus since? Have you learned anything? Is there a way we can use Libratus to improve our poker skills? Would CMU offer a course?
fsgfaf (1 год назад)
Why did these idiots agree to these conditions? Obviously without being able to take breaks, not having rest days, not having a dinner break, having to take the time to study AFTER they have played all those hours each day, they are going to play worse than if the conditions were better. Impossible to say how much worse they played and what the outcome would have been if they were fresh. Did they not learn anything from Deep Blue and Watson, both of which were set up to advantage the computer over the human players?
RoiAleTea (1 год назад)
wait wait wait wait 14:50 explains that this test was complete nonsence. They aren't even playing poker if they are equity chopping all in scenarios, folding becomes a much larger mistake, jamming with draws becomes way more profitable, and river mistakes are an enormous hit. I can't believe these guys aren't loudly exposing this flaw in the test.
jiwei snow (1 год назад)
hi doug, have you played poker snowie bot? If yes, what general leaks does it have? thanks in advance, it crushes me HUNL
Zsolt Bihary (1 год назад)
Im wondering... The bot just plays the Nash optimal strategy, or does it try to adopt to its opponent, to extract maximum value?
Zsolt Bihary (1 год назад)
So whats up with this challenge? I hope you will make a summary once it finishes.
kickalion (1 год назад)
your jacket is like 2 sizes too big for you
Brad Stollery (1 год назад)
So Doug, what does Libratus' (fairly comprehensive) victory over some of the best NLHE players that our species has to offer say about the state of AI regarding HUNL? Has this specific variant of poker now been "solved" if humans can no longer beat a machine?
Dane Nexo (1 год назад)
Doug, the quality of your content is phenomenal but the speed at which your quality increases every video is beyond outstanding. It's commendable how you've just been able to produce such high quality and quantity of video and do it right; this obviously reflects on how you became a successful (arguably one of the best NLHE players in the game {even still}) poker players there is by your obvious work rate. If theres anything anyone can learn from you and translate to a poker skill - HARD fucking WORK. All the best Doug and keep going.
yztyzt1 (1 год назад)
it is a damn bot. It should have been done at the human's leisure. The human should log in and play whenever and forget about the mirror. The mirror didn't limit variance anyways because each street was different. It didn't matter what the hole cards were. God damn idiots at carneigie mellon fukcs.
dachevashe (1 год назад)
I wonder if whether the bot team uses hands folded by human players as learning data is a strong factor in "fairness" of this challenge. What do you think? Is it stated somewhere in the rules? I think for the development of poker AI it would be wasteful to throw away that data, but at the same time it looks very unfair to use it _during_ the actual challenge.
Jonathan Gray (1 год назад)
I'm wondering what blinds and stack sizes they were playing against Libratus. All the references to win/loss amounts will be easier to think about knowing this.
PETAR LAMBREV (1 год назад)
obviously only human left that can beat AI at HU is Dan Bilzerian - he plays nowhere near to logic
Sean Martinez (1 год назад)
IDK if anyone else has said this already, but with regards to the equity chop, I would make the case that it decreases variance human vs. computer, but it increases the variance of results among players (comparing human vs other humans.) In the AJ vs A9 example, you get it in on the turn, you win 95% of the pot. You get it in on the river, you lose 100% of the pot. Without equity chop, both players would suffer the same result of losing the entire pot, so they would share the same result between them, but their variance against the computer would result in higher swings. As far as the Libratus coders care about, they would probably prioritize reducing variance between human vs. computer, and not care as much about the variance of results among players. So while both perspectives have merit in truth, the equity chop has the more relevant result to see if humans can beat this AI over the long run.
Sean Martinez (1 год назад)
RoiAleTea what I mean is results of the individual players vs Labrontis as opposed to all players' results compared to each other vs Labrontis. so among these 4 players, theyre all playing mirrored hands in order to reduce the variance of the results of the experiment from one player to another. so an equity chop would actually increase the variance on that stat depending on when the players get the money in. if one player shoves turn with 95% equity, and another is all in on the river with 0%, the difference between their results vs Labrontis are going to be very different rather than if there were no equity chop, they both lose the hand and have the same result in score/money. however, if both get it in on the turn, then of course that reduces variance against the computer. it's just an issue of when players play their hands differently, they're going to get different results from each other, so equity chops would increase that difference rather than level it.
Sean Martinez (1 год назад)
RoiAleTea what I mean is results of the individual players vs Labrontis as opposed to all players' results compared to each other vs Labrontis. so among these 4 players, theyre all playing mirrored hands in order to reduce the variance of the results of the experiment from one player to another. so an equity chop would actually increase the variance on that stat depending on when the players get the money in. if one player shoves turn with 95% equity, and another is all in on the river with 0%, the difference between their results vs Labrontis are going to be very different rather than if there were no equity chop, they both lose the hand and have the same result in score/money. however, if both get it in on the turn, then of course that reduces variance against the computer. it's just an issue of when players play their hands differently, they're going to get different results from each other, so equity chops would increase that difference rather than level it.
RoiAleTea (1 год назад)
I don't understand how weather it's HvH or HvC makes a difference. In Kim's scenario, the computer won 100% when it had A9 because the hand went to the river, and when it had AJ it also won 95% because they got it in on the turn and never saw the river. Maybe I'm reading your comment wrong but aren't you contradicting yourself saying "it decreases variance human vs. computer" and then "their variance against the computer would result in higher swings"
Demondoink (1 год назад)
this is so motivating to improve. can't wait to grind tomorrow!
fuzzybaffy (1 год назад)
So... what does this mean for professional poker? Will professional poker players be out of jobs now?
Fedor Amuricano (1 год назад)
WTF, just 54k suscribers ! common guys... Sub sub sub
Christopher Graff (1 год назад)
Did I miss a Cookie Story ?
Eahab Ibrahim (1 год назад)
its a different game when you can't get psychological tells from a bot. Humans however can give away tells, and the AI can get some sense of them. They could even in the future start training them for emotion recognition
Isaac Lau (1 год назад)
Hey Doug, quick question: what are all these guys online nicknames? Thanks for the videos btw keep it up.. They've been absolutely awesome
PokerMan (1 год назад)
You know one of the kids from the school has this shit running on ignition as we speak from his house... lol
Jon H (1 год назад)
Doug I feel like the equity chop combined with the mirrored hands tends towards favoring river bluffs vs. turn bluffs. One of the guys mentions the 53 suited that flatted a 4 bet pre and taking the aggressive spot on the river, I wonder if there are many more examples of this type of play from the bot? What are your thoughts?
Jon H (1 год назад)
Essentially, any turn bluff can be rebluffed or the bot could just get it in to realize any possible equity, where the river is the only spot (obviously) where the equity is pure. Since the humans are so far behind the bot can really apply more pressure to deny equity because the humans need to see more rivers to maximize winnings to reduce the computers lead. Or maybe I'm not seeing the full picture!
Gary Gwinn (1 год назад)
These AI developers, are stealing the data from the human players, to improve the BOT.
Loriene Lozano (1 год назад)
great video
J Tran (1 год назад)
Lol I feel funny that a player of Dong Kim caliber still can't grasp the concept of variance. He shows ONE case where equity chop in a mirrored match increases the variance. What about the other thousands cases?
McGavel1 (1 год назад)
Awesome info - thanks!
shutemdwn (1 год назад)
Ok, how does everyone know this bot doesn't know all the hole cards?
Vincent Borgmans (1 год назад)
Close that top button, bro!
Rynorulz117 (1 год назад)
Wow Doug, nice moves. Keep it up, proud of you.
dan mizalino (1 год назад)
thank you doug polk for all you do for the poker community you are a true poker genius keep it up bro
Dobrin Ivanov (1 год назад)
hahah dan polk :D :D: D:D ....... btw some sections of people talking are put twice in the video, still it was like 10 seconds not more
Doug Polk Poker (1 год назад)
+dan mizalino Appreciated, from a fellow Dan
Dylan T (1 год назад)
It's elementary that equity chopping reduces variance. I see exactly what you're saying but it's an extremely naive view. It's pure mathematics, equity chopping reduces variance. I'm pretty surprised you and Dong are confused by this incredibly simple concept.
RoiAleTea (1 год назад)
@Dint xoxo and Pluto is a planet. Stay scientific Jerry
Dylan T (1 год назад)
No, I'm not overlooking those scenarios. You're still wrong.
Robert Szasz (1 год назад)
In normal situation its obv that it reduces variance, but you overlook that hands are mirrored here. So this is a totally different situation and they have a good point that actually increases variance. You can see that by example they give, but for more concrete stuff: so if you have AAvsKK in a spot then in the mirrored spot you have KKvsAA if you both get it AI equity chopping does nothing with variance (with eq chop you get 80% 20% split and mirrored case 20% 80% it means both get a full pot or basically nothing vs each other, without eq chopping you get 100% with one hand and 0 with other and in the mirrored spot the reverse of it so its the same as before, but if one of the guys slowplays KK and calls (which strategy mostly should have the same EV as jamming preflop otherwise you should always call) and the flop comes like Kxx and they stackoff the KK gets like 90% of the pot while if in the reveste situation the other guy went AI preflop gets 20%, if the AA slowplays its the same result) and obv there is variance involved too when you get postflop etc... but by simple example you can see with mirrored hands equity chopping most likely really increase variance or at least that's it not so trivial that it reduces it by any means.
Robert Mihai (1 год назад)
that tiny tiny Mic tho :) :)
Joe Dybisz (1 год назад)
Lmfao Dong is my favorite. They all seem like chill dudes though
super soch (1 год назад)
doug go to a gym, ur getting fat!
all bdo (1 год назад)
Great content Doug! Been fascinated by AI vs Humans since Kasparov vs Deep Blue in 97 and what followed in Watson with Jeopardy, AlphaGo, and your coverage of the recent AI vs Humans HU matches.
Brad R (1 год назад)
If you continue to have trouble beating the AI, it may be time to use pokers secret weapon. Introduce it to the DONK, and its almost certain that the AI will find no way to outplay the runner-runner 2 outers.
Fred Fenner (1 год назад)
Great vid but I'm so confused about the equity chopping thing. In Dong's story, he said Jimmy got it all in on the turn but they never see a river? Scratching my head.
Robert Szasz (1 год назад)
Okay, then i didn't understand well what you didn't understand :) Equity chopping simply just means chopping a pot in proportion to the equity each players hand has. For example if you get it in preflop AAvsKK the player with AA gets 82% of the pot and the one with the KK gets 18%. In the example w A9 vs AJ on A9xJ the A9 has 5% eq and the AJ 95%. And yes " all-in the player who is behind is returned a portion of the pot based on the percentage that they are behind before the board is completed" this is exactly as you said.
Pedro Marte (1 год назад)
equity chopping means that nothing is dealt after allin, you get the % you have, so if they get allin on turn A9 gets 5% and AJ gets 95%, if allin is on river A9 gets 100% and AJ 0%, huge difference, equity chopping redcues variance only if used alone, the combination of equity chopping with paired hands is what makes variance increase. its funny because both things alone reduce variance but when they are mixed TOGETHER they increase variance
Stephan Altena (1 год назад)
If you get it all in on the flop, you get whatever percentage equity you have, there is no turn en river. In this case A9 only had 2 outs. So he only gets 4-5 % of the pot. Because no river is dealt.
Fred Fenner (1 год назад)
...or could it be that when there is an all-in the player who is behind is returned a portion of the pot based on the percentage that they are behind before the board is completed? Because that seems like it would change elements of your strategy, no?
Fred Fenner (1 год назад)
Robert Szasz Buddy I'm not unfocused, there's just incomplete information here. In the vid they do a whole interview segment about equity sharing without explaining what it is; I'm trying to use the context to figure it out. Maybe explain what is meant by "Dan gets 5% of the pot back". If he has the same runout, he wins the all-in, so what is there to "get back?" Do you maybe mean he returns 5% of the pot? And how does this reduce variance?
Cabluigi (1 год назад)
Super interesting video. Hope humans turn it around!!
Johnnie Ho (1 год назад)
Have you thought about reviewing any hands from the challenge for a Poker Hands video?
Jay-Jay Okocha (1 год назад)
Am I the only one who noticed that the interviewer dude looks like Dan Polk?
Doug Polk Poker (1 год назад)
+Ben 86 I hate people that look like Dan
Dan The Jester (1 год назад)
Love that vape naysh music
Kaspars Burka (1 год назад)
Video series reviewing most interesting hands of the matched would be great
mcpartridgeboy (1 год назад)
2 words .... deep blue !!!!
salens (1 год назад)
Can you post some info on the guys playing the challenge? What stakes do they play? How good are they? I'm asking since they're not known. Also, how much money did they get for this? I'm guessing they aren't going through this madness for free food.
CribNotes (1 год назад)
Does the bot talk smack at the table?
Martin Grysbaek (1 год назад)
You just sat down with Jason Lease without asking for permission to come aboard? Rude Kappa
Doug Polk Poker (1 год назад)
+Martin Grysbaek Haha wp
Cafe Con Leche (1 год назад)
Jason Les can't even handle a little speech play. We need Will Kassouf to step up for team humans.
dru713281 (1 год назад)
You fit the middle aged fat coach perfectly those man boobs are pointy and epic.........and your still running from me I see .......100/200 game..........guess sitting behind ur keyboard and playing online is plus ev.....and playing against the best in the world is negative ev.....
Guilherme Bernardes (1 год назад)
yeah online players are the worst,,, lol
dru713281 (1 год назад)
+mcpartridgeboy and I don't wanna hear excuses about who is in the game or this or that a true pro goes after the best.....doug wants to be a celebrity now and I cant hate that when you expose your strategy your giving opponents the formula that will destroy you over time................
dru713281 (1 год назад)
+mcpartridgeboy oh I just speak the truth bro no need for lessons from a guy who pros salivate over when he is in the game. there is a reason he is never gonna win player of the year and continue to charge for lessons when other pros prove by their work rate....I have been playing the game for 16 years and let me tell you there is a big reason doug stays online and away from the sharks.....he is the fish in those games.
mcpartridgeboy (1 год назад)
great tactic, but it doesnt work, i tried insulting chess players on youtube to get a free lesson from them by annoying them so much with arrogance they would give me a game, wich i would obviously lose ! but it never happened ! nice try though !
sean19922k7 (1 год назад)
Finally a more realistic microphone size.. can't deal with the high standards set these days
Thorsten Brunner (1 год назад)
nice interview doug, your doing a great job
Doug Polk Poker (1 год назад)
+Thorsten Brunner Thanks!
Stephan Szilvasi (1 год назад)
I love McAulay. "Fucking gangster"
SirSvavarsson (1 год назад)
so are we watching the end of days for online poker,... rip
DionisOdam (1 год назад)
Awesome video DAN POLK! I have some thoughts I'd like to share and ask you about. It's obviously clear that the bot doesn't get tired while humans do. Why then the ones who established the rules of the challenge didn't just extend the time in order to give the humans the possibility to compete at their pick? At the end what it matters is just the number of hands played and not how much time needed, right? All the players were saying the same! This should definitely be taken into account since it's very likely that machines are going to make better and better than us in the future. I honestly don't think this is equal and impartial as a scientific experiment should be.
J Crush (1 год назад)
hello Doug....I want to watch Human vs AI poker where do I go or where do I have to look for?
kingenidjingeln (1 год назад)
It's pretty obvious that the A.I will be stronger then the human player in the future. But why rush it and sponsor such a thing? Can't we enjoy internet poker w/o getting tilted by bots. Shiet, im pretty sure they can develop a bot for some good use instead of developing it for online cheaters. Because that is what it's going to be used for in the near future.
HardRock Whisker (1 год назад)
Wow, so if the AI wins, does that mean that poker is obsolete? Online poker will be dead for sure, but how it will affect live poker? Will it have a chance to survive after that?
M Ef (1 год назад)
You didn't mention why you arent playing this time Doug, care to elaborate for us? Its gotta play the best or we'll never really know..
M Ef (1 год назад)
Thanks Doug, very limited content online, legend!
TheTomtah (1 год назад)
Maybe review the last hand of the 2017 assie millions main event from today for next "poker hands"? Pretty sick call for all the biscuits in the end. I thought the call was too loose but turned out to be a good call. What do you think ? You can find the vod on jcarverpokers twitch channel.
Christopher Eklund (1 год назад)
Mitchell Reid (1 год назад)
I have a phd in statistics. I can see what you're saying with the equity chops thing but you're wrong. Even in the mirrored context equity chops reduce variance
smaakmakend (1 год назад)
wow this really conserns me if the bots slightly own you then they demolish any stake up to 50nl hu on stars what does this mean for the future of poker.. what if this shit gets into the game like it did with videogames?? maybe pokerstars can afford to keep them out but the rest wont have the money....
Ant G (1 год назад)
the conditions these pros are playing in are so poor. it sounds so unfair. they should implement fatigue into AI. I play HU every day for several hours and it is very demanding.
` hExQuisite (1 год назад)
I agree with lots of people that they should be treated much better, I mean, separate to sessions to relax, sometimes 1 day off etc. Because this way is just marketing, we all know that it is different, because the bot beats 4 completely exhausted players, however people and the news will only mention that the AI won... Just for the fun, I would be also interested if we would mix, an educated weak reg, and a completely unpredictable LAG fish. I don't say here they would perform better, I'm just curious how the AI would adapt to a guy who knows what he is doing, but still with several leaks, and a guy who is a complete maniac recreational (Ban Dilzerian?)
Sean Gallagher (1 год назад)
Hey Doug cute microphone
kindergitarre (1 год назад)
And I thought Dan Polk is a poker expert, meanwhile he is just a reporter.. maybe his bilzerian story is also made up and he isnt a fraud!
produstef (1 год назад)
the empire, aka embassy cause that's one poker ambassador right there. very comprehensive and comprehensible poker content. keep up the good work mr. polk
Li Hsing Chen (1 год назад)
Doug, do you feel it's still worth putting all the effort in poker while AI has become this strong even beating top 15 HUNL pros?
1282craig (1 год назад)
well doug, 2 seconds in and i'm already chuckling. well done 'danny-polker-hands'
Paul Archer (1 год назад)
Doug, I believe you have the same bit in here twice at 2:08 and around 13:15. Don't know if that was intentional, just looking out.
Christopher Hannel (1 год назад)
Doug would you be able to provide a clear explanation of the equity chop. I didnt quite understand what you guys were talking about
po1980 (1 год назад)
I have a background in cognitive psychology (spent 9 years in phd program, have refereed pubs, etc). I think the crux of the problem is this- let's take an extreme example where the humans are forced to play the bot for 500 days in a row, 11 hours per day, no breaks. the bot wins. So what?!?! obviously no one would be surprised by this result. so at what point is the game "fair" for the humans? the researchers are obviously aware of this. that's why they made the study have no dinner breaks, grueling hours, for many many days in a row, etc. the players must be allowed to rest, regenerate their mental resources, take breaks, eat, etc, in order for this "game" to be "fair".
philhellmuff (1 год назад)
Is the bot team tweaking it during the challenge like between sessions or are they hands off and it's just on auto pilot and adjusting using AI? Would be cool to see if they just let it loose on stars for a month to see how it does and then return the money.
Demondoink (1 год назад)
i think if it beat these guys then it would destroy the players on Stars. no way Stars would EVER let a bot play on their site willingly in a game for $.
Kamino (1 год назад)
Why didn't you get a real microphone? :D
Humberto Teixeira (1 год назад)
"Sounds romantic..." Doug knows about those "bed talks"...
Humberto Teixeira (1 год назад)
Doug struggles so much to get that scottish accent! hahaha
Lachie Panetta (1 год назад)
you can't say nlhe heads up is solved unless it beats someone like Doug, jungle, ike haxton
Cade Murphy (1 год назад)
Doug i never comment on YouTube videos, but I have to tell you how much I appreciate your content. It's awesome keep up the good work. I only play casually now but I'm hooked to your videos. New subscriber here.
Ernie Acosta (1 год назад)
Terrible double vid set up.
chadmichael03 (1 год назад)
Doug, I saw the notification of this new video in my inbox, and I thought I was about to see a political video
Lars Lestadius (1 год назад)
Need subtitles because of Daniel
Jordan Holiman (1 год назад)
"youre all great players, and i know, because i coached all of you". hahaha. nice one. good video very interesting
Catend (1 год назад)
I'm taking an AI class and play a bit of poker so I'm very interested in this coverage. I hope there is more to come!
phoshigs (1 год назад)
Vape Naysh yall!
Hugo (1 год назад)
good stuff man
Stephen Jones (1 год назад)
Yo that ending credits clip of coach Polk was hilarious.
Mason Barrell (1 год назад)
Will the hand histories be made public after the challenge?
FearMilk (1 год назад)
Nice vid douglas. cant say im surprised, AI is the future. just bought me some shares in brainchip ltd
Harbz (1 год назад)
We are all doomed! Save us Dan, save us!
cochiselol1232 (1 год назад)
Can we all just take a minute to thank Doug for putting in so much effort to keep his viewers happy!
cochiselol1232 (1 год назад)
*Dan Polk sorry
Transcendent (1 год назад)
Keep in mind, even if the bot wins, it likely won't be by a statistically significant margin - e.g. it will be a "statistical tie".
Ryan (1 год назад)
The end boss doug polk! Fantastic content here my man. These players gave us some good insight. Thanks for making the trip and interviewing these guys and bringing this to us.
Andrew Morgan (1 год назад)
Doug, thanks for all the vids, great way to past the time
Clabonne Maric (1 год назад)
Great video !
Everytime (1 год назад)
Doug don't slack on the lifting!
xBuGGzBuNNyx (1 год назад)
Only common denominator I see that's different is Doug Polk. Doug Polk is the answer here. Doug Polkk made these guys. Doug Polk >>>>> Everything, Everyone and Anything
Steven Hir (1 год назад)
The interviewer holding the microphone did a great job
Arnold Layne (1 год назад)
yeah he held that mic really well
UPRlSlNGG (1 год назад)
Dan Polk woulda smashed this bot.

Хотите оставить комментарий?

Присоединитесь к YouTube, или войдите, если вы уже зарегистрированы.