AI Revolution in Go: Corners & Sente

[@vastabyss6496]

The AlphaGo documentary is actually what introduced me to Go and made me pursue AI research as a career. This is such a cool game, thank you for all the lessons!

[@GoMagic]

Thank you for watching!

[@armentsirunyan6937]

Me too! KZbin suggested this documentary about a game I had never even heard of before, and everyone seemed so in love with the game that I had to give it a go!

[@jeffb1886]

The documentary is really wonderful...whether you play Go or not (I do!). As exciting as DeepMind is for the growth of AI and neural network tech, one couldn't help but root for the human (Lee) and feel like we've all lost something that WAS (but is no longer) uniquely in the domain of the human brain. Game 2/Move 37 was the moment it all changed. I was as crushed as he was. 😞

[@JadeCrystal-so5ib]

Hi, I'm from Brazil, thank you for everything. Unfortunately here almost nobody appreciate this magical game😮😅

[@GoMagic]

Hello from afar, and thank you for your kind words! 😊 It’s true that Go isn’t as widely known in some places, but that’s totally fine-you can be the one to start a Go club in Brazil! You might be surprised at how many people will become interested once they see how magical the game is. Every Go community starts somewhere.

[@angelogunther6445]

I've recently taken up GO with a friend and your tutorials have been invaluable in teaching us everything there is about GO, Thank you so much. Sadly GO does not seem to be widely played in Europe or the US. Are there any good online GO apps you can reconmend for practice?

[@hosguitar]

Ogs is a good website that have very active go communities and resources like joseki and tsumego

[@gabri9728]

If you want to play online I suggest OGS, Pandanet or KGS. Tsumego Pro does a great job for tsumegos. Also BadukPop is very nice

[@monsignorjames]

There's oneline-go, you can play games online, though I find the puzzle section to be too hard, even after knowing the answer.

[@GoMagic]

Hey. We're happy that our content helps. As for apps... First of all, we have our platform. There are a lot of interactive courses out there: gomagic.org/course-categories/ If you're not interested in courses, you can solve some Go problems in our Skill Tree (which is available even without paid membership): gomagic.org/go-problems/ And in case you just want to play online, read this article about Go servers: gomagic.org/online-go-servers/

[@angelogunther6445]

@@GoMagic thank you very much for the information 😊

[@nomukun1138]

The "best move" is only the best if you have perfect execution at every point of the game, and many AI-determined "best moves" seem to thrust the game into a state of confusion no human can understand. Are there AI tools that have a focus on explaining the state of the game in human terms, instead of "this move is 78%, this move is 79%." ?

[@GoMagic]

Nope, there are no such tools atm

[@_timelike]

Even in alpha go, these percentages are not just numbers. They are based on a search tree, and it's not so hard to let AI to show those most probable trees (or let AI show the variations it is thinking) in for any engine really. It just a matter of who will invest the time to build this tool

[@madcroc111]

@@_timelike KaTrain already is a free tool with AI stronger than any human and it shows the moves AI is planning to do. Very helpful for learning, though still very difficult to figure out exactly what AI is thinking.

[@klauspoetsch1841]

Great and simple explanation!!

[@GoMagic]

It's great that you found our video useful 😊

[@orang7525]

Just wanna know what kind of stones you are using? And where can I get them?

[@GoMagic]

These are Shin Kwang Baduk Korean glass stones (Jang Stone II). 8mm Bi-Convex. I guess you could try to find them on Amazon.

[@Мопс_001]

AI doesn't think in the same categories as us humans. It can take a very risky path that is worth only and only in case of perfect play. So when we try to play more AI-like moves that supposedly lead to an advantage we might not know that this advantage is of a highest precision and can be ruined by one imperfection. We make moves that give us bigger chances of winning by making our position better so there is less room for a blunder. Would be cool if someone made a hybrid evaluation bar that synthesizes the winning chance by combining the perfect play AND proportion of those moves that will work to those that won't. Now THIS bar and highlighted moves would be cool to play with since this system would be more sensible and give human-like advantage that we can use.

[@GoMagic]

Yeah, that sounds really interesting.

[@NIMPAK1]

Nowadays there's a lot of fear when it comes to AI, but I feel that the AI advances in Chess and Go prove that AI can be a helpful tool, especially when it comes to learning. However I do think that there's the downside where AI could potentially take away the discovery aspect especially for top players and Go could potentially become a game with no experimentation.

[@GoMagic]

I guess you're onto something there. But I don't believe in people abandoning experimentation altogether. The board is too big, the move are too many.

[@kohismahpimp]

AI has proven that the closed perspective of any game is meaningless. If we want the game to remain interesting, our perspective and thinking about the game has to change. The reason why we enjoy playing games was never the goal of completing the game, but how the journey unpacking something hard for us, changes us. To that end, you can always add discovery to the game, by simply adjusting your constraint or what you value. Isn't it interesting that chess has evolved with different formats? Isn't it interesting that something like poker, which has always had a known algorithm maximizing statistical outcomes, is still played and watched? Afterall, at the very least, you can always decide that losing most of the time is fine if at least you can win an upset sometimes, and decide to take risks in uncharted territories. This, is what makes professional play interesting for an audience!

[@hvk224]

Excuse me, I'm recording later. There is a section in the video where you explain that we cannot afford to ignore these steps. 4:40. Personally, I think that something you've gone through to improve your senses in a certain period of time will begin to feel that's fine. They're used to knowing how to ignore things we didn't get along well, I'm sorry and he came after you, but it would have occurred to me to say your things in the video

[@GoMagic]

Thanks for sharing your thoughts. It's true that over time, players develop a better sense of what can be ignored and what demands attention. And we’re glad you’re finding the videos helpful. 😊

[@SD-bv5gc]

I really enjoy Calculations games Is Go game at the level of chess in this case?

[@GoMagic]

I'd even say the Go is a level above. Just imagine the number of move variations. 🧐

[@SD-bv5gc]

​@GoMagic Are you saying that go has more calculations than chess to predict the next moves?! If so, that's really great because in chess a lot of calculations are done to predict the next moves !

[@Pyth445]

​@@GoMagic you sure go has more complicated calculations for anticipating moves than chess?

[@GoMagic]

Yup. Please note that I'm not attempting to claim that Go is superior to Chess. It simply demands more strategic foresight and the sheer number of potential moves at any given moment in the game... well, it's staggering.

[@prestonwoods8197]

@@Pyth445you should watch alphago documentary

[@Pyth445]

I heard that go has balance of strategy and tactics however, chess is more tactical and less strategic Is it because of the size of the board? Because I guess in the smaller board in go, tactics are more than strategies .

[@GoMagic]

Yes, Go board is very large and players need to think through how events in one part of it will affect another part.

[@SD-bv5gc]

it's interesting! But is it all about board size? I know that a bigger board size definitely increases the strategy in go and chess but Except this, not possible that the fast movements of the chess pieces (except the pawns) reduce the strategy in chess ? And the slow movements of the stones in go increase the strategy in go? And if this is not true (And only the size of the board affects the strategy and tactics in go and chess) does it mean that strategy increases as the size of the chess board increases? (and chess will be a balance of strategy and tactics) and does strategy decrease as the size of the go board decreases? (and go becomes more tactical and less strategic)

[@GoMagic]

It's not an easy question tbh. Wall of text incoming! Yes, the size of the board plays a significant role in both Go and (probably) in chess. I'm not sure about chess though, since we play chess on the same size board. But probably a larger board in chess would mean more squares and potentially more complexity, as each piece can move across more possibilities. In Go, a larger board (such as the traditional 19x19) also increases the complexity, as there are more intersections where stones can be placed, leading to a richer array of strategies. A smaller board simplifies the game, reducing the depth of strategic planning needed and possibly increasing the tactical interactions, as each move has a more immediate impact on the game's outcome. As for the piece movement... Probably. Fast movement of some pieces in chess create more opportunities for swift strikes. This can sometimes emphasize tactics over longer-term strategy. In contrast, in Go stones are placed permanently, and such slow, incremental placement emphasizes a more strategic approach. So I'd say you're right. Both the size of the board and the "speed" of the pieces affect the balance between strategy and tactics. However, the board sizes for both of these games are standard, and the pieces move differently, so we have what we have.

[@SD-bv5gc]

​@GoMagic thanks ! that was great information Actually, there is infinite chess now(infinite board) and this makes me think that The depth of Go game on the infinite board will be unimaginable! ( Getting points with the finite parts of the board that we have taken and getting points from the capturing opponent's stones The game ends when one of the players reaches the number of points chosen by both sides(naviary)) I hope one day infinite go will be made This is one of the things that must be experienced before death !

[@Py-445]

​@GoMagic Hi Sorry for open this again I did a lot of research on this and finally came up with two question : 1_What do you mean by the fast movements of the chess pieces cause swift strike and that Increases tactics and reduces the strategy in chess ? I have a problem with the concept of fast and slow here Does this mean that this will cause an immediate result in the game and the game will end quickly, and long plans cannot be designed and for these reasons, tactics become more than strategies in chess? (And how is this case on bigger boards with more chess pieces?) Or do you mean the fast movements of the pieces because they are dynamic and very quickly, naturally increase tactics and decrease strategies in chess ? (So in this case it doesn't matter anymore that even if on bigger boards the game is very long and has no immediate result and you can design very long plans because in all of them the pieces are dynamic and fast and that's why the chess game is more tactical and less strategy) Well, which of these two is correct?(Or maybe neither or both?) 2_Can you almost say how many percentages of tactics and strategies are there in the game of go and on the 9×9, 13×13, and 19×19 or infinite board ? I think in the infinite board, the game is still a balance of strategy and tactics, because the stones are still slow and placed permanently, and because the game ends with the Choice of points by two players, and if the points be chosen little, the game is short and with Immediate result without long plans and if the points are chosen a lot, I think the game will be a balance of strategy and tactics Strategy because of stones placed permanently and game is long and long-term plans and tactics because of stones can be placed anywhere and local fights and short plans and.... so what do you think? Please use the percentage Because I know that in bigger boards the game is a balance of strategy and tactics and in smaller boards is more tactical and less strategy I searched a lot for find answers to these questions, But i did not reach a result And since you gave a very complete answer last time, I said it is better to ask you these questions Thank you if you can answer 🙏

[@myboy_]

Wow, the ai in the thumbnail is so good it even got four moves in a row!

[@reebaanee8586]

It's handicap 😂

[@GoMagic]

Phew, thanks god you have noticed. We will immediately punish this pesky AI for cheating. 👮‍♂️ Please stay on your guard and keep analyzing all the AI arts in our videos. We're counting on you! 🤗

[@myboy_]

@@GoMagic the shade is amazing 👏

[@suntzu6122]

Lol juuuuuuudge!

[@BtrDaysAreComing]

Hmm... Many things to say. First, is this really about sente, or about the value of taking a corner? Second, the best move against AI is not always best against humans. An usually move (a trick move) is effective against humans. If you play only AI moves in the opening you lose the surprise effect, the ability to bring your opponent into uncharted territory and keep the initiative.

[@GoMagic]

Yup, that's right. But even so, you'll have a chance to surprise your opponent in the midgame.