Alpha Zero's "Immortal Zugzwang Game" against Stockfish

[@FrenchComprehensibleInput]

Stockfish: e4 AlphaZero: looks like an interesting mate in 100 puzzle

[@antikokalis]

lololololol

[@FrenchComprehensibleInput]

@@mr_clean575 Crazy how things have evolved in a couple of decades!

[@joescott8877]

@@FrenchComprehensibleInput We ain't seen nuthin' yet. Wait till a month from next Tuesday, LOL.

[@rmendeljacobs2832]

Agadmator on his death bed: _croaks_ "He-hello everyone. I think this might be my last day on this planet but that's no excuse not to show you my daily chess video"

[@user-uj2cq6rd8n]

lol

[@juggernaut316]

lmao

[@أفلايتدبرون-ث7غ]

Not but he always say butt 😂😂

[@MrJoyDevision]

He’s an excellent KZbinr

[@Splozy]

"Hello everyone, I am currently dead, but there's no reason not to show you this amazing game from the other side"

[@ZanzibarBreeze]

Now Stockfish understands how we feel when we play it

[@edgarallanpoe295]

Zanzibar Breeze xD!!!

[@stagna1959]

Not exactly. Even after zugzwang move g4 it does not understand that position is completely lost.Every GM and even patzer knows it, but stupid computer keeps playing because it "valuates", "Oh, I have rook vs bishop and two extra pawns, it cant be so bad " .It even does not consider that black queen is trapped and completely useless POS even though every 1400 rated beginner knows it . Rybka valuated position after g4 as 0,5 . GM would resign right after g4.

[@DavenH]

Reporting this stolen comment to google

[@stagna1959]

What comment ? Stolen from who ?!

[@DavenH]

The first one in this thread was on a different AlphaZero vs stockfish video. I wasn't being serious btw

[@Rspknlikeab0ssxd]

Stockfish: The position is drawn. Alpha Zero: ThE pOsItIoN iS dRaWn

[@isawicameiconqueredandcame3708]

Lmao

[@dirty7444]

Alpha Zero: Hold my power supply....

[@cobi617]

@Audi a4/s4 WrITinG In ThiS ManNeR mEaNs theY ArE MocKinG YoU

[@Fiskie666]

@@cobi617 Thanks for this interpretation.. it still seems not fitting at all. Also I was trying to analyse the capital letters if there had been any meaning to it .. then it might have been somewhat of a .. clever reply? Like this it is more like .. okay I changed the way the characters look .. and that is funny .. No.

[@cobi617]

@@Fiskie666 Who said i was trying to be funny, i just wanted to explain with an example

[@RBG9000]

Interesting to see an engine give up the queen in frustration xD

[@brunesi]

Yes, I do find that quite interesting, for it should know it was a loss right? Usually it would resign, of did it not see it coming?

[@TechnetiumPrime]

@@brunesi stockfish is incapable of resigning

[@mixedvibes9613]

@@TechnetiumPrime actually? It makes sense but fr?

[@Brakum]

@@TechnetiumPrime except that this video ended with him saying that stockfish resigned

[@Brakum]

@@mixedvibes9613 no because if you watched the video you would know tbat stockfish DID resign

[@nathanjxaxson]

*Holy moly,* that qeenlock was mindbustingly brilliant. That is just unreal. I have never seen a human create such a decisive shutdown of a queen without actually trapping/capturing it.

[@futuropasado]

No, the queenlock is a mistake by Stockfish that had 4 better moves and choose that one, game was rigged. This is marketing, stockfish wasn't playing it's better version. Check an engine, tha game was equal till that exact moment Qh8. Bad move.

[@jordanlu9152]

Azku Shang stfu stockfishy boi

[@repaleonhalo9754]

@@futuropasado it's weird why he would lock it into a corner

[@CallOn84]

@@futuropasado actually, during the time it was played, Stockfish 9 was the latest. As Agadmater pointed it out, the only difference between the rating was one or two hundred difference.

[@vedrangrubac1849]

Watch Andersen's immortal

[@SimonTeSlime]

"I think I'm coming down with the flu... my head hurts and I'm shivering a bit but that's no reason not to show you another match" Love your dedication!

[@davidegallo2185]

In 2020 this comment has another tremendous meaning

[@suhailad4834]

@@davidegallo2185 may we all be safe!

[@samtakhanna2497]

Yes

[@demerzel3798]

Best Chess bots: 3. Stockfish 2. AlphaZero 1. Lip stick Engine

[@zzcc229]

What about Agadmator- my initial plan was to resign

[@cocolee7644]

Exactly LIP STICK ENGINE plays better than you LOLOLoLoOlOloLoLlOlOloLoLOooOLOooOoOlOlOl

[@almscurium]

To cheat you oxygen thieving dingleberry

[@nxchz698]

No it’s not, it’s 3.stockfish 2. Alpha zero 1. Eric Rosen “I’m low on time”

[@demerzel3798]

Akash Del Actually I think Eric Rosen “oh no!” is stronger

[@joebobtnvs2521]

14:12 "After Stockfish gives up the queen anyone can defeat it" You underestimate my power

[@chrisallen9509]

Honestly Queen and bishop vs two rooks at 3300 strength doesn't seem too easy. Definitely disagree with agadmator here but that's cause he's a 2000

[@parzingtheasian]

actually, he vastly overestimated my power, i couldn't win if it was a king vs my three queens

[@bc_7644]

@@chrisallen9509 bruh you dont need to be a pro to win that endgame, alpha is up the exchange with an unstoppable passed pawn and stocfishes king can't do anything.

[@dark_rit]

I think I can handle that endgame in that position. Like sure someone completely new to chess would still probably get crushed if they took over in this position, but someone who has played chess before should be able to win from there given the gross advantage white has. Like I put the boardstate into an analysis board and white's almost +4. It's really hard for black to make any moves like the passed pawn is basically untouchable unless black wants to get checkmated sooner, but even with top engine moves only that the computer tells me black is checkmated on move 72 when pawn to a4 was move 60. Black's f pawn is just way too weak and has to be babysat, but it still doesn't matter because you can defend the f pawn or you can double up on the a file to stop the passed pawn you can't do both.

[@fuckyoutube4398]

@@chrisallen9509 Yeah, if it were just those two piece vs 2 pieces. But Queen/bishop has a passed pawn, rooks have to always keep ONE rook at the back flank/have them connected, because mate is threatened by two pawns that are difficult to reach (again, without being able to pressure with 2 rooks at once) and all the queen needs to do is hit that back rank for checkmate. You plug this in and Stockfish will probably put white at like +10 because anyone with even 1800 rating can win this.

[@jpray22]

so my boss just caught me watching this at work and made me explain myself... him: are you playing online chess at work? me: ughhh... no him: wait... youtube? are you watching someone playing chess? me: well, not exactly. him: who is that in the corner? is he playing? me: no, hes analyzing a game between two computer programs... him: so to be clear... you are watching a guy, watching two computers play chess, while you're at work? me: y-yes. him: *sighs, walks away*

[@agadmator]

+Josh Pray I laughed so hard :D I shared this on all my social networks, hope you don't mind :D

[@MikeCNC]

Top Comment

[@moran22226]

Your boss could perfectly be replaced by Alphazero

[@vladislavzedano3145]

hahahha

[@dragonsdraughts8382]

top tier laugh.sometimes there is gold in these youtube comments.

[@Souvik_Dutta]

10:08 Stockfish immediately grabs it. He says this is bad for white . Alpha Zero: Call an ambulance but not for me.

[@cemluca1624]

Pulls out a gun

[@happyboi347]

28 stab wound

[@niklace]

Queen to H8 was weird, it defends nothing with the queen. Would a human do this move?

[@jro3213]

Call the amber lamps!

[@drewpaiva93]

It defends checkmate on the dark squares@@niklace

[@GutsToCuts]

“Remember everyone, it’s not a good idea to move the same piece more than once in your chess openings” AlphaZero: “haha, you silly humans and your arbitrary limitations”

[@superman5742]

As if stockfish wasn't already depressing enough 😭

[@Motivationtoescape]

so my boss just caught me watching this at work and made me explain myself... him: are you playing online chess at work? me: ughhh... no him: wait... youtube? are you watching someone playing chess? me: well, not exactly. him: who is that in the corner? is he playing? me: no, hes analyzing a game between two computer programs... him: so to be clear... you are watching a guy, watching two computers play chess, while you're at work? me: y-yes. him: sighs, walks away

[@null_pointer_deref]

@@Motivationtoescape Congratulations on copying the most liked comment of this video

[@silentoccasion4359]

@@Motivationtoescape goddamn plagiarizer of comments

[@dashnaso]

@@Motivationtoescape so my boss just caught me watching this at work and made me explain myself... him: are you playing online chess at work? me: ughhh... no him: wait... youtube? are you watching someone playing chess? me: well, not exactly. him: who is that in the corner? is he playing? me: no, hes analyzing a game between two computer programs... him: so to be clear... you are watching a guy, watching two computers play chess, while you're at work? me: y-yes. him: sighs, walks away

[@trevors3908]

God, I love hearing "Stockfish resigned the game."

[@benc6190]

it really says something when stockfish thinks its a draw and then it RESIGNS just moves later!

[@fivethreeone2132]

. .

[@lukaskuipers7791]

@@fivethreeone2132 why do you exist

[@fivethreeone2132]

@@lukaskuipers7791 im gay

[@introverteddawg9805]

@@fivethreeone2132 ok

[@brunesi]

@@lukaskuipers7791 Alpha zero is not managing our genetic pool yet.

[@davidegallo2185]

4:28 You can't move the same piece twice AlphaZero: Haha, the queen goes boom bing bang

[@xj1471]

AlphaZero was xQc all along

[@davidegallo2185]

@@xj1471 Must be hard to be a streamer, a chess engine and also Hikaru's house ceiling

[@AlwaysAudacity]

Oh man. Can you imagine if A0 refutes Stockfish's refutation of Morphy etc. ?? What if Danny Rensch's CAPS is shown to be wrong by A0, and the old masters were the best? What an exciting time for chess!

[@apollon011]

Well the machines actually do play better chess as A0 is not human as far as I know. So maybe the question is not, can the AIs play a Morphy move and then close out the rest of the match (as agadmator posed it in the video)? The question might be: can very advanced AIs like A0 (rated 4000+?) 'see' so deeply that they can pull off the 'magic' of Romantic chess and play Morphy-esque moves without any problem? Can they refute the late 19th/early 20th century's refutation of the Romantic style of chess? Maybe Romantic chess will experience a sort of revival thanks to these AIs. And then maybe in the late 2020's super advanced AIs will counter-revive super deep positional chess to refute present-day A0's refutation of present-day positional chess. And then things will come full circle! Who knows? The sky seems to be the limit with the AIs of the future...

[@rezganger]

Rensch is a clown!

[@juliansoto2651]

It seems that A0 has an amazing heuristic method that can analize complex closed positions. Most of chess engines are bad at playing closed positions, I've NEVER seen a chess engine putting its opponent in zugzwang, it is so surprising for me, i wonder if it can solve hard chess puzzles, like this www.chess.com/forum/view/general/a-chess-problem-that-computers-can-t-solve

[@kasparov937]

Julián Soto Its incredible.

[@liamwatts5267]

Sigmar the old style romantic chess was only made possible through opening mistakes and bad play by the defenders (e.g. to greedy, not knowing when to give back material or just outright blunders).

[@rahulsuthar8590]

I don't comment on youtube videos but when I do ; I am really impressed. And this time by deep mind alpha zero. Impact of Impression by deep mind is like tal Impression . Just can't ignore it

[@tensor1129]

Rahul Suthar we're honored really

[@fritagonia]

Really agree with you. Amazing

[@CodeProvider]

these AI games are incredible. it's interesting how they play while following most chess theory, but like you said Alpha ignores the common wisdom about developing all your pieces early. very cool

[@BenjaminJLadd]

Yeah, watching A0 games will end up reteaching us basic strategy. We'll know for sure what works and not just have a general idea of winning strategies. That's pretty cool.

[@kkoromsp3674]

Instead of developing our pieces, we should de-develop the enemy pieces.

[@knightoflambda]

Funny enough, that's how it went in AlphaGo's games against pro Go players. It likes limiting its opponent's potential.

[@Flaxify]

Well to be fair the bishop was on its default square nevertheless it was developed as it controlled the right squares

[@Mrtre7]

Code Provider, Benjaming Ladd. I don't see this at all. What I see by white is multiple moves with the same piece in the opening (even, the queen). I see rooks crammed on the 6th row without the files being open. I see non sequiturs like Rd1 c4, Rd5 (when possibly A0 was first worried about Qe1+, then c4 changed some 30-moves-deep tree and made the threat negligible) and finally a winning combination no human mind would be able to even imagine. Where is the instructional value, "the basic principles" and "chess theory" applied here?

[@djr36djr36]

Running Stockfish 12, three years later, it does find 44. Bb3 if you let it run to depth 34 or so. It also takes quite a while to find 46. Qb4 but gets there in the end. It thinks its big mistake was at 49.. Rf8. Its evaluation flips from +1.4 to +6.6 at depth 30. It thinks it should have played Kf8 instead, giving his queen a square to help defend the f7 pawn, which frees up the e8 rook. The move it still can't find is 47. Rxc5. At depth 42, its top three suggestions are Qf4, Rd6 and Qh4, all of which it rates as basically draws (+0.4, +0.2, +0.1). But, show it Rxc5, and it very quickly recognises that White is improved (+1.2).

[@Marcusjnmc]

thanks

[@iliillillilli2991]

Running stockfish 14.1 now, it sees all these moves immediately at around depth 24. It’s amazing to see how much it improved now.

[@niranjanrajesh1058]

@@iliillillilli2991 yea depth 24 is where it sees the rook sac. Unbelievable that around 18-20 moves in it doesnt even consider the rook sac as one of the best moves. Its chilling to see the eval fall to 0.0 after the rook sac and then suddenly go back up

[@TaranVH]

Incredible.

[@ivanng2014]

Unexpected Taran

[@genericusername4206]

who r u

[@Elogamer15]

playing chess with macros??

[@mozisi]

Stockfish: "It most likely looks like a draw" Alpha Zero (in Aronian's tone): "Maybe that's why I'm a great player and you're not" On a serious note, I refrain from using profanity in the comments but Holy Fuck that was beautiful.

[@brettluther7303]

lol.. Good one

[@bradc3402]

And even better it not only gives the suggested move that it claims will draw, it then says the move that alpha zero makes is bad, and then it is just utterly punished trying to show why! I love the fact like all engines it refuses to just resign, and must be embarrassed by having to sac the queen for absolutely no compensation!

[@tedceldor]

Aronian to Maurice: That's why i'm a great player and you're not. Be Finegold to You Tube viewers: That's why i'm a grandmaster and you're not. Ha ha ha ! What a joke. Seriously Alpha is not AI it's ET.

[@nobody27019]

Alpha Zero games are like a drug. I don't know how I will deal with the abstinence after you are done posting these games.

[@jamesfleming1155]

Nobody agreed lol

[@milz7129]

James Fleming nobody agreed? I agreed. Lol his user

[@NaderTaghinia]

It’s just so enjoyable for me when Agadmator gives a half smile and shakes his head in admiration of a move...love it 😀😀👍🏻🌿

[@sirhamalot8651]

I've watched other commentaries that were 45-50 min long! THANK you for being brief and to the point. A great run-down of the game.

[@stevenwilson5556]

This is one of the few channels that I would be happy to set to "auto-thumbs up" on every single video. This channel sets the standard on chess channels, and really on all sports and entertainment channels that analyze any performance by athletes or competitors. Very deep, thoughtful, and respectful of the game and it's players. Bravo.

[@VijayThakurMD]

While being funny and entertaining.

[@Joa8n]

I agree🙂that's because of agadmator's abilities..he's not only a good commentator but has a very satisfying voice and other abilities which I don't know how to explain😂it's like watching a new movie. There are many many chess commentators but agad is the best😍😍😍

[@VijayThakurMD]

@@Joa8n and agads audience is surprisingly good, and full of mature folks.

[@hydra9267]

Thank you so much for covering these games by Alpha Zero as it completely opened my eyes to some new possibilities in the fields of learning and excelling at things, as well as reinforced some beliefs i had for quite some time.For years i'v thought that learning something by yourself from the basics, starting just with it's rules and principles and getting a deep profound understanding of it is the absolute best way to learn.To explore and add/subtract by trials and tribulation rather then blindly follow the preexisting formulas and methods.After hearing about Alpha Zero's learning curve and seeing a couple of it's games gave me strong confidence that it's true.For years our brains have been polluted with unnecessary information about this or that even before we have learned the core structure of it, while if we approached everything new as pure and played with it like a child we would have a much better understanding of it in the end.The way this Morphyesque "engine" plays chess was just a delight to watch.It not only seems to make the best sequence of moves but also the prettiest :).It seems like we as a society have excelled so far that we started to take things for granted and forgot that some of the very best in their fields like Tesla, Da Vinci, Morphy just to name a few - were self taught and we needed an AI like Deep Mind to reminds what really matters. And damn this comment was tough to write in English, pozdrav iz Srbije :)

[@Radjehuty]

This is probably my favorite game of the A0 series. It still blows my mind how it managed to achieve such a ridiculous position. It inspired me to play the game again and I love following Leela in her amazing TCEC matches. It really is calling in to question how much weight we should be putting in to chess engine strategies. Leela has just recently played in the Benoni opening as white and seemingly destroyed SF10. mind blowing.

[@rohangeorge712]

same this is my favorite alpha 0 game, beatiful immortal game against another AI

[@leroyjenkinson4866]

9:43 I let stockfish run for a couple hours and at depth 53 it actually recognizes the bishop move as winning for white (+0.9)

[@JoicSeth]

Depth 53?? That's insane

[@samcroswell2386]

And I am assuming this is with 3 years of development for Stockfish? I suppose it shows Alpha was just ahead in development during this match. Its promising that Stockfish that is available for us to analyse our games are continuing to develop so far.

[@argschrecklich9704]

@@samcroswell2386 Alpha Zero was given the rules to Chess and it trained for a few hours playing against itself. It wasn't ahead in "development". It was originally created to play Go. The difference between alpha zero and Stockfish is that Alpha Zero is a machine learning neural network, and Stockfish is basically a sorting algorithm.

[@samcroswell2386]

@@argschrecklich9704 Oh I see. I guess 'development' is the wrong word to use then. You can't really say 'talent' because its an AI, but what I mean was that Alpha Zero was ahead of Stockfish in the 'understanding' (I guess) of chess by years haha. Its a bit surreal talking about an AI like this though.

[@argschrecklich9704]

@@samcroswell2386 It really is surreal. I feel this is no longer just a "chess algorithm", it's a "chess entity". But the most interesting aspect is that it's practically able to master any perfect-knowledge game within hours simply by been given the basic rules and then playing against itself. It's the world's best player in Chess, Go and Shogi (and protein folding, which is very important in medicine). The main difference to Stockfish and similar engines is: those have no real understanding of chess. They simply go through all possible moves ("brute-forcing"), pruning the nonsensical ones and keeping the rest which they evaluate through human-coded criteria and sort them accordingly and then pick the ones with the highest value. It's a "machine" in the purest sense of the word. AlphaZero is more organic. It looks at far fewer moves per second than Stockfish, - just like humans. But its "experience" let's it focus on the best ones more easily - just like humans. We can't really say what's going on inside a neural network (just like with humans lol), but AlphaZero seems to show traits and also a playstyle that look from the outside like planning, having ideas, being deliberate. Nothing of that was taught by humans, it's self taught, and it took only 4 hours for it from being switched on to beating Stockfish. Stockfishs weakness is basically that it was coded by humans, with human preconceptions in mind. We can, for example, observe that AlphaZero has less emphasis on piece value, it values a good position more; because it was not taught by humans, so to speak, it's not weighted down by our human understanding of the game, it learned the game as if it was the first entity to ever play it, fresh and from scratch. This all might sound boring in theory, but it's actually a really exiting development and the next decade will be absolutely wild in the area of AI, machine learning and neural networks.

[@uberneanderthal]

10:38 even the latest version of stockfish doesn't initially recommend this rook move and calls it a mistake, until it goes into deeper analysis and at depth 32 says it's the top move with +1.2

[@simonemiglioli1165]

Bishop move. The rook comes after.

[@DocsDota]

I think one of the most overlooked ways of playing chess as most humans have been taught was trapping pawns from pushing. We've emphasized most of the chess strategy with using pawns to push and pieces to support them Alpha Zero's approach was to put pieces on top of them to immobilize them from entering the game. The amount of grounds gained without actually opening up the board was quite interesting of a new concept.

[@michailmichailidis2530]

I just wish Tal, Nezhy, Spielmann, Anderssen and all the other ''romantic'' attackers were here to witness this. I have always believed in attacking and I will continue to do so. Thank you Google Deep Mind!

[@harsh3608]

I completely agree....maybe the "romantic style" of playing chess is really the way to go and chess is all about finding the right ideas and combinations and not just playing boring positional games that ultimately lead to draws if played "perfectly" by both sides

[@gheffz]

Fantastic game ... great game! I like the way Alpha Zero shook Stockfish around like a rage!

[@Edzianni]

This game is amazing! It is not only beautiful, it really shows how AlphaZero approaches the game of chess with completely different ideas from those of conventional chess engines handcrafted with human knowledge.

[@sevendayoptions6704]

Wow, watching alpha zero play is more entertaining than 99% of what's on tv these days. Awesome match!

[@AshishTiwari-mz7yl]

The way agadmator is showing this games it looks like there is some world championship going on between this two..and google deep blue took the title of best chess gamer in the world from stockfish

[@aracaniusinfinius2880]

Ashish Tiwari its more something along the lines of A0 butterslapping Stockfish and saying "I don't like material i like activity" and then upgrades the butterslaps to bitchslaps

[@aracaniusinfinius2880]

Ashish Tiwari its more something along the lines of A0 butterslapping Stockfish and saying "I don't like material i like activity" and then upgrades the butterslaps to bitchslaps

[@MrSupernova111]

Its historic. Much more important than some match. Whats even more interesting is how AI will affect other fields. Im in finance and we should be scared as AI can virtually do all of our functions if trained properly.

[@jesusthroughmary]

That is essentially exactly what happened. Stockfish's world champion title now rings hollow.

[@fad1969]

+jesusthroughmary Without its opening book and endgame database, Stockfish was essentially crippled.

[@asdfmovies831]

Please make more alpha zero games

[@dimimo1972]

There have to be more released to the public. FYI, these are the 10 games. You may find the article interesting too. cdn.chess24.com/GzFl-Z4-SVWO-mC9rL6XhQ/original/mastering-chess-and-shogi-by-self-play.pdf

[@antonioviriatoferreira3641]

Man, that was awsome. "make more alpha zero games"...Lol

[@aravinthbalakrishnan3427]

Deep mind is absolutely mind-boggling, it seems like a very human playing style. Reminds me of morphy

[@OneDerscoreOneder]

Aravinth Balakrishnan how do you know if something looks like it's playing like a human

[@izs6946]

Alpha zero considers activity over material, but that's always a risky game. I have never encountered a computer engine that thinks like that

[@OneDerscoreOneder]

is that primarily just because it's hard to make an engine that "understands" activity and is able to value it in the right way? Or just because it's a new type of neural net that's special in some way or something

[@MoiMagnus1er]

I would say it is link to the fact that A0 learn by itself. So contrary to Stockfish that just search for the most efficient strategy, A0 also search for the most efficient way to learn chess. Meaning that he will mainly find the strategies that are "easy to learn through experience" and not strategies that are "easy to learn trough mathematical analysys". That's why its way of playing is "more human" than Stockfish's way of playing.

[@Randomguy-wd5lw]

human and AO are neural network that why, but stockfish use more simple tradditional programming

[@brucknerian9664]

Your channel is a great improvement over most others who cover similar games but rush through the games like they need to go put out a house fire. Thanks for this more in-depth relaxed approach. You deserve an academy award or something similar.

[@anujgupta3397]

"And stockfish resigned the game" said no human ever Alphazero- Hold my beer

[@-zelda-]

Engines don't resign

[@mohammedjafer9265]

@@-zelda- they do

[@-zelda-]

@@mohammedjafer9265 No, uci engines are literally incapable to resign

[@adityabhandare]

More more alpha zero gamessss. Thanks

[@leakedzebra]

Is it just me or does anybody else love Alpha Zero's games a lot more than human games? I do love all the high intensity and time pressure games, but still these are just beautiful, in a very Picasso-esque way

[@boskovich]

They should start a game after some controversial Tal move and let A0 refute the refutation :)

[@rokeley94]

I really love seeing the games between chess engines. Thanks for all your work!

[@jonathanhutton5100]

I really enjoy this channel a lot. I was drawn to these videos because, as a Go player, I have followed the development of Google's Deep Mind programs, marveling at them along the way. I love the way agadmator describes the games and the reverence he shows for the program. It's incredible that after only 4 hours AlphaZero is able to play such an elegant and beautiful Chess game. Much like its Go game its Chess game never seems brutish or vulgar... it flows over the boards like water and dances circles around the greatest of opponents. It's truly something to behold.

[@brendanward2991]

Now I know how the Europeans felt in 1858, when Paul Morphy turned the game upside down.

[@TheHigherSpace]

Alpha zero plays like Tal hhh in the way that he always sacrifices and empowers activity over pieces ... But the entanglements and the trapping of pieces that can't move that A0 creates are unmatched ... it's just amazing watching this ...

[@arvindhmani06]

It felt like the perfect piece of art watching it come together

[@mclaughlinja1995]

This game is somehow both beautiful and sort of terrifying at the same time.

[@niranjanindia8278]

This is the Best game I've seen in all my life..thank you.

[@royalemaster3713]

Agadmator: "I think I'm coming down with several tropical diseases that require immediate medical attention but that's no excuse not to show this beautiful game"

[@asdfmovies831]

Thanks agadmator for the beautiful game by the.... Terminator chess engine

[@giannisgougoulias3443]

LOL

[@nickshtenikov1261]

T1000 against T800

[@Pintkonan]

@@nickshtenikov1261 stockfish is nowhere near being a t-model series. its one of those dogs barking at the bunker doors when a t-model tries to enter.

[@alexcerullo3143]

Pintkonan are you dumb? StockFish is the best chess engine in the world (technically Alpha Zero isn’t an engine)

[@markkealy4417]

@@alexcerullo3143 why isn't it an engine?

[@TheSonicSegaNerd]

Stockfish: The position is drawn AlphaZero: Well yes, but actually no

[@liamhalliday8437]

I saw this game nearly 2 years ago, and it came back on my suggested feed today. I recognised the clip immediately. I had to watch it again. It's still beautiful, and the fact Stockfish never saw it coming just makes it even better to me. I think this game was the best in that original series, it's just awesome. Before AZ I found computer chess dull, a grind I couldn't understand. Now I find it rekindling my love for aggressive, attacking chess, and playing in a way that I feel I should be able to as a human, if only I were a super genius.

[@Marek-gn9jn]

"And the Queen can never enter the game again" litle shaking with his head... I swear I felt like a kid when I was bullied out of sandbox once :D You made my day agadmator :)

[@rohanp]

Thanks Antonio, I could never have interpreted the game on my own . You explained it so well.

[@stevebuhrt3504]

"How I became AlphaZero's Victim" The new chess bestseller by Stockfish, Mark Taimanov Approved XD

[@samtakhanna2497]

Lmao

[@Zarko02]

It's absolutely incredible this engine I hope we can see some more video

[@xDMrGarrison]

I watched this a while ago when I was just getting into chess, and now that I've played thousands of games I can appreciate it much more. This is insane what Alpha Zero did, absolutely mindblowing :D And again STOCKFISH no less, a freakin' 3400 rated chess monster.

[@Akash_Tyagi_93]

Antonio enjoyed the humiliation that Stockfish suffered. 🤣🤣👌 And Agad's intro is as good as the finish Alpha ZERO provided to the game.

[@Zekbo]

It’s truly incredible to have the chance to view a game which has been played by a mind far more intelligent than any human ever born. Just to witness this transcendent intelligence at work is beautiful.

[@tomasseeber]

The best thing about these games are the uncountable horrible variations poor Stockfish was able to avoid.

[@nathantonye2465]

Show alpha zero as black

[@LorenzoStroe]

Alpha is too alpha to play as black

[@GeronimoStilton93]

White: Stockfish Black: AlphaZero 1. e4 e5 2. Nf3 Nc6 3. Bb5 Nf6 4. d3 Bc5 5. Bxc6 dxc6 6. 0-0 Nd7 7. Nbd2 0-0 8. Qe1 f6 9. Nc4 Rf7 10. a4 Bf8 11. Kh1 Nc5 12. a5 Ne6 13. Ncxe5 fxe5 14. Nxe5 Rf6 15. Ng4 Rf7 16. Ne5 Re7 17. a6 c5 18. f4 Qe8 19. axb7 Bxb7 20. Qa5 Nd4 21. Qc3 Re6 22. Be3 Rb6 23. Nc4 Rb4 24. b3 a5 25. Rxa5 Rxa5 26. Nxa5 Ba6 27. Bxd4 Rxd4 28. Nc4 Rd8 29. g3 h6 30. Qa5 Bc8 31. Qxc7 Bh3 32. Rg1 Rd7 33. Qe5 Qxe5 34. Nxe5 Ra7 35. Nc4 g5 36. Rc1 Bg7 37. Ne5 Ra8 38. Nf3 Bb2 39. Rb1 Bc3 40. Ng1 Bd7 41. Ne2 Bd2 42. Rd1 Be3 43. Kg2 Bg4 44. Re1 Bd2 45. Rf1 Ra2 46. h3 Bxe2 47. Rf2 Bxf4 48. Rxe2 Be5 49. Rf2 Kg7 50. g4 Bd4 51. Re2 Kf6 52. e5+ Bxe5 53. Kf3 Ra1 54. Rf2 Re1 55. Kg2+ Bf4 56. c3 Rc1 57. d4 Rxc3 58. dxc5 Rxc5 59. b4 Rc3 60. h4 Ke5 61. hxg5 hxg5 62. Re2+ Kf6 63. Kf2 Be5 64. Ra2 Rc4 65. Ra6+ Ke7 66. Ra5 Ke6 67. Ra6+ Bd6 0-1

[@GeronimoStilton93]

White: Stockfish Black: AlphaZero 1. e4 e5 2. Nf3 Nc6 3. Bb5 Nf6 4. d3 Bc5 5. Bxc6 dxc6 6. 0-0 Nd7 7. c3 0-0 8. d4 Bd6 9. Bg5 Qe8 10. Re1 f6 11. Bh4 Qf7 12. Nbd2 a5 13. Bg3 Re8 14. Qc2 Nf8 15. c4 c5 16. d5 b6 17. Nh4 g6 18. Nhf3 Bd7 19. Rad1 Re7 20. h3 Qg7 21. Qc3 Rae8 22. a3 h6 23. Bh4 Rf7 24. Bg3 Rfe7 25. Bh4 Rf7 26. Bg3 a4 27. Kh1 Rfe7 28. Bh4 Rf7 29. Bg3 Rfe7 30. Bh4 g5 31. Bg3 Ng6 32. Nf1 Rf7 33. Ne3 Ne7 34. Qd3 h5 35. h4 Nc8 36. Re2 g4 37. Nd2 Qh7 38. Kg1 Bf8 39. Nb1 Nd6 40. Nc3 Bh6 41. Rf1 Ra8 42. Kh2 Kf8 43. Kg1 Qg6 44. f4 gxf3 45. Rxf3 Bxe3+ 46. Rfxe3 Ke7 47. Be1 Qh7 48. Rg3 Rg7 49. Rxg7+ Qxg7 50. Re3 Rg8 51. Rg3 Qh8 52. Nb1 Rxg3 53. Bxg3 Qh6 54. Nd2 Bg4 55. Kh2 Kd7 56. b3 axb3 57. Nxb3 Qg6 58. Nd2 Bd1 59. Nf3 Ba4 60. Nd2 Ke7 61. Bf2 Qg4 62. Qf3 Bd1 63. Qxg4 Bxg4 64. a4 Nb7 65. Nb1 Na5 66. Be3 Nxc4 67. Bc1 Bd7 68. Nc3 c6 69. Kg1 cxd5 70. exd5 Bf5 71. Kf2 Nd6 72. Be3 Ne4+ 73. Nxe4 Bxe4 74. a5 bxa5 75. Bxc5+ Kd7 76. d6 Bf5 77. Ba3 Kc6 78. Ke1 Kd5 79. Kd2 Ke4 80. Bb2 Kf4 81. Bc1 Kg3 82. Ke2 a4 83. Kf1 Kxh4 84. Kf2 Kg4 85. Ba3 Bd7 86. Bc1 Kf5 87. Ke3 Ke6 0-1

[@GeronimoStilton93]

All the games published ==> arxiv.org/pdf/1712.01815.pdf

[@villaholland]

nathan tonye he won 3 with black 25 with white lost none and drew the rest

[@iamverynicesir]

Alphazero seems to be the new hope for the chess community. Makes me wonder if Tal might've given these engines a harder time, or if Kasparov's loss was just a little bit too overhyped and perhaps these engines aren't completely unbeatable. You play like an engine against an engine you will lose everytime. Play with some creativity though, some romance? Maybe you'll do better than the grandmasters of our time say you will.

[@BeatPoet67]

Tal would have been crushed. He made a lot of unsound sacrifices which would have been instantly refuted by these machines. But these machines will crush any human.

[@iamverynicesir]

But in this very video we see the AI make an "unsound" move according to Stockfish, (the latest version of Stockfish too since it was agadmators doing the analysis) and it did the crushing. These engines are not perfect. The only people claiming they are are the same players who do nothing but get destroyed by them and then copy their tactics and keep getting destroyed by them. Tal did not play a lot of engine approved moves in some of his most famous games and yet the opposition would be destroyed. I wasn't saying he would do as well as the AI did in its games, but with his unique style he might have gotten draws and maybe even some wins from time to time.

[@jefffokarat27]

Your right, Human brain(mind) is more than AI. 👍

[@Todestuete]

I get your point, but overall humans just can't keep up anymore. Some unique style or out of the box moves doesn't help it. Deep Mind might prove that current engines actually play chess wrong, but they still play good enough that they crush any human even if he tries to apply "Deep Mind tactics". Against handicapped current engines with an Elo not much higher than that of current GMs, a different play style might work for GMs to beat them.

[@couchpotatoe91]

If with "romance" you mean "analyzing 70,000,000 positions per second then yes...

[@groussac]

"Favoring activity over material." A good way to put it. Thanks for showing us this update on artificial intelligence.

[@petergreen5337]

Beautiful lesson and demonstration thank you very much publisher

[@fep_ptcp883]

Hahaha Stockfish seems like lost its patience in the end!

[@bezzlebedeviled4756]

9:59 -- The Stockfish10 engine (newer that the one used is the video) recommends Bb3 here as one of several 0.00 drawing choices (with White having no better moves). 10:22 -- Stockfish rates Qh8 as the best move for quite awhile (e.g., still 0.00 out as far as depth 33). It has trouble finding the Rxc5 line after the Qb4, Nc5 followup because there are so many other lines to explore than don't immediately go down material. Even after Qh8, Qb4, Nc5, it doesn't begin considering Rxc5 until depth 29, but still rates it no better than 0.00 out to depth 33 -- because the considered line does not forfeit capturing the pawn in lieu of Qh4. After Rxc5, bxc5, then and only then does it quickly find Qh4, and pump White's rating. So, all in all, Qc7 should be played instead of Qh8. 11:11 -- Rf8 is the real blunder that costs Black the game (Stockfish10 almost immediately discards it from consideration at over +6 in White's facor). Kf8, followed by Qg8, protect the f7 pawn. Play from there continues: Qc7, c4, Qxc4, Rd8, Qd4, a5, g4, Kd8, Ra6, Qe8 (Black is inchworming back to the center attempting to unlock his queen), g5 (protecting the attacked h6 pawn), Re1 (the queen can breathe, but must guard h8). The rating here is about +2 for White (but only +1 if he immediately captures Black's a-pawn) , with all lines exchanging into drawing positions where White is usually up material but cannot capitalize. One interesting example continues (after Re1): Bc4, a4, Bb5, Re6, Bxd7+, Rxd7, Ke7 (both queens are staring a rook in the face), and White must either commence perpetual check, or draw a queen versus two rooks endgame after Rxf8, Kxf8).

[@quonomonna8126]

Alpha Zero proves that you win on solid principles, not on thinking 20 moves ahead

[@MrSupernova111]

Revisiting this old gem! Thanks!

[@CNCTEMATIC]

Its fascinating to watch Alpha dance with the pieces and see Black's position getting more and more constricted through moves it is forced to play.

[@ΓιώργοςΝανούρης-β3σ]

This machine will bring a new era

[@bruceli9094]

Alpha Zero has the souls of Casablanca, Tal, Morphy & Fischer.

[@owendavidmalicsi5900]

David Bronstein: "The most powerful weapon in chess is to have the next move" Stockfish: "yeah! right?

[@JW-bo3ql]

Thanks so much for the work you're doing on these games. If you do more, I'll watch them all!

[@kirkstable]

This is still my favorite game that you've ever covered thanks you're on my Christmas card list this year

[@chrysfredi2883]

Zugzwang is german and means to be forced to make a turn

[@Vivungisport]

Its that moment when you aren't chasing for the initiative.

[@D_scxnnect]

the intro in this video is absolutely terrifying in 2020 lmfao

[@Yas9in]

Unfortunately it is the same in 2021

[@birthdayzrock1426]

What's so terrifying about the flu?

[@mamounjo]

I've never seen chess played like this before. Bizarre.

[@gorilaz0n]

This’s so exciting! You guys should have a joint review with someone who’s both good at go and chess! It’ll be fascinating to see how AlphaZero transfers it’s learning style to chess. I won’t be too surprised if it’s queen moves in the opening resembles to its forcing moves in go opening. I’m a 5dan in go but I’m shitty at chess. I can only imagine how much it can help my go if I understand how it’s learning style in chess too!

[@einarjuel]

This is a historic video. Thanks, Agad :)

[@Savage-ws7sy]

Alphazero: Does a machine like yourself ever experience fear!?!

[@PaleGhost69]

Quote is my life.

[@r2out]

That's an interesting proposition. The implanted AI for the memory part of the game, and the ancient human brain cortex for creativity.

[@stefan0ro]

b.b.but stupidity has a role in spiritual evolution

[@r2out]

Would you care to elaborate on that? What role has stupidity in spiritual evolution?

[@marianpalko2531]

PaleGhost69 Neuralink is on the way...

[@antonioviriatoferreira3641]

Lol!!!

[@Nytequiller]

That's crazy how stockfish thinks it's a draw and 2 moves later the game is over

[@eskay6666]

Thats the most impressive part. Its a draw, then two moves later hes like FUCK! I have to sac my queen

[@illusion6559]

I first watched this video 1 year ago today and its what got me started in chess. Just wanted to say thank you :)

[@vadimmedvedev7895]

I rarely watch a chess video more than once. This one is an easy exception. Just a beautiful game with great commentary of course.

[@mitchstilborn]

These Alpha Zero games are freaking amazing. Keep em coming! :-)

[@BakedPotatoYT1]

Bad thing is, the future of chess could be engines vs. engines, the one with the strongest engine wins or even some engine tournaments.

[@agdjdh4689]

Chess is infinite for human brain. In every game there will always be new moves.

[@sigurd106]

There are already engine tournaments?

[@haywoodtew8090]

"Once it's given up its queen anybody can beat it" I don't think you've seen me play an endgame.

[@bra-steve]

is it just me or does alpha zer make a lot of queen moves...maybe Aronian was wrong

[@fadhlihsani]

Lonwabo Mvovo maybe it does know that queen power must be used at maximum potential early on.

[@bra-steve]

Fadhli Ihsani so Lev was wrong in the interview then😂

[@AGriffith]

Lonwabo Mvovo No,he was right, rember that Levon said that there is a rule that you have to move more than one time your pieces...

[@bra-steve]

Anthony oh I thought he said there is a rule that you shouldn't move a piece more than once 👍

[@MatsMatsuo]

Deep mind uses it's queen forcing black on not doing development moves and improving the queen's position, very interesting

[@pajarillo2723]

The way Alpha Zero craps the enemy position is insane. That rook move at 6:24 stunned me for a while.

[@hoehlentroll8284]

Taking the queen off the board without taking it... mindblowing brilliant.

[@prakhardwivedi3649]

Upload all 10 games!!!!!!!!!!!!!

[@midmid6]

Hey, could you maybe put up an analyzation from stockfish about the position? That way we sort of 'know' when stockfish is thinking it is going to win :) And when it realises it is doomed ;)

[@jamesfleming1155]

midmid6 this is such a good idea. But it could do it yourself.

[@DarianCabot]

Here after Demis Hassabis mentioned this video to Lex Fridman

[@AriasFco]

Wonderful video mate. Good handling of the subject, passionate and insightful commentary.

[@moundhirmoundhirs6123]

I never played chess...thanks to you i started to do so

[@markangus3252]

Welcome to the chess world. The chess world itself isn't a beautiful place but chess itself has incomparable beauty.

[@StarzzLAB]

Alpha zero is way more stronger than any chess engine because it's knowledge about chess emerges only from simple chess rules and playing with itself. This is no some brute force BS like Stockfish or Houdini.

[@martybishop6131]

I just now finally understood what alpha zero really is...... predator (the same one from 1987) came from space hunting human chess players from the past and the present not for their skull but for their brains to get our revenge of the current chess engines that ruined our romantic chess.. Its a deadly combination of Kasparov, Fischer,Karpov, Carlsen,Capablanca,Morphy,Tal,Lasker,Botvinik,Alekhine and all the other great players made into ONE bad ass mother fucker chess wizard!!!!!

[@rapbeats402]

If you analyze this game with Stockfish it gives 49. ...Rf8?? a blunder and says better is Kf8 upon which it gives white only a 0.6+ advantage, the idea being if Alpha Zero carried on with its plan of g4 there is Rc8 and the King escapes to the queenside and black can hold to a draw. It really seems suspicious to me that Stockfish rated over 3000 can make 3 inaccuracies, 4 mistakes and 2 blunders in a game.

[@Ohdeerhere]

i mean, from what i've heard it wasn't the latest version of stockfish, was played on an inferior machine, and both players were only given 1 minute to move. did you let it analyze the position for longer than a minute? probably.

[@gmoves4482]

Yeah everything’s a bit sketchy

[@youssefm6061]

The two has the se time, and stockfish analized 100 millons vs 400.000 of alpha zero. Not excuse

[@spinninglink]

I agree with the op. The newest stockfish found the Rf8 blunder under 5 seconds. Seriously try it yourself. After 10 seconds it shows Rf8 losing by more than 5 points. With the Kf8 move, after more than 30 minutes of calculating, it shows white with just a +0.36 advantage. I don't think alpha zero would have won without that huge blunder by this older stockfish version.

[@yonkouterkuat4835]

What make A0 great is, its learned for only 4 hours If the latest version of SF can beat it, A0 can just learned for 5hours or 10hours more and become more powerful isn't it?

[@hermancharlesserrano1489]

Amazing, Alpha Zero really does bring the swashbuckling back to Chess

[@nagavignesh9292]

I got goosebumps when he said ," here it is , the immortal zugzwang