AlphaGo - The Movie | Full Documentary

It's intensely contemplative It's almost hypnotic It's like putting your hand in the third rail of the universe If you play seriously there's a chance you'll be exposed to this experience that's like nothing else on the planet It's putting you in a place where you're always in the loop limit your ability, there is a reason why people have been playing Go for thousands and thousands of years. It's not just that they want to understand, they want to understand what understanding is and maybe that's really what it means to be human I was a kid I loved playing games I started with board games like chess and then I bought my first computer when I was eight years old and I won a chess tournament since then I felt that computers were this kind of magical device that could expand the power of your mind.

We believe that virtual environments and games are the perfect platform to develop and test AI algorithms. Games are very convenient because many of them have scores, so it is very easy to measure them incrementally. progress so I'm going to show you some videos of the agent system, the artificial intelligence, so let's start with the escape, so here you control the baton ball and you are trying to break through this rainbow colored wall, the agent system you have to learn everything. on its own just from the raw pixels it doesn't know what it's controlling it doesn't even know what the goal of the game is now the start after 100 games um you can see the agent isn't very good he's missing the ball most of the time time, but it's time to get the idea that the bat should go to the ball now, after 300 games, he's as good as any human being can play this and he pretty much gets the ball back every time we think, well, that's great, but he left the system playing for another 200 games and did something incredible: he discovered that the optimal strategy was to dig a tunnel around the side and put the ball behind the wall.

Researchers are working on this amazing AI developer, but no. very good at outbreaks and they didn't know about that strategy, so they learned some of their own system, which is, you know, quite fun and quite instructive. I think about the potential of general AI, so for us, what's the next step now is the most complex game ever devised by a man. Beating a professional player at Go is a long-standing major challenge for AI research. When I was 18, I won't change my life. That's why I go to French. I won't try to forget. the goal but it is impossible because all the things I learned in my life is this visco looks like a mirror I see the goal I also see myself for me gold is a real life dear mr pham my name is demyster harbis I have an artificial intelligence company based in london called deepmind as the strongest go player in europe we would like to invite you to our offices in london to meet you in person and share with you an exciting go project we are working on if you are interested come visit us please let us know know.

Thank you. Best regards, Demis. When I see this email, I don't know if it is true or not. I will accept this. Why not? For me, Iris is an adventure. I want to go visit Deepmind to find out. What is this? Let's go to the project. The first visit. I think maybe he wants me to sit in the special room. I push a lot. Many use it in my hand. way into the deep mind to see that we were a serious operation and we were serious people, we are doing proper research and the search time is getting better and better, as we go, for example, from one to eight in any of these, always increases.

Think of deep mind as a kind of Apollo effort for AI, our mission is to fundamentally understand intelligence and recreate it artificially and then once we've done that we feel like we can use that technology to help society solve all kinds of other problems. If we look at the actual game, we can see what

alphago

thinks, what is the most likely variation that he thinks will happen. We've been working on

alphago

, our gaming program, for just under two years. Now all the little patterns cascade together. After layer, after layer, after layer, I started talking about the game of go with demis over 20 years ago, so this has been a really long journey.

The game of go is the holy grail of artificial intelligence for many years, people have looked at this game. and they've thought, wow, this is too hard. Everything we've tried in AI just falls apart when you try the game of go, and so it feels like a real litmus test of progress if we can figure out go, we know. We did something special, so in the rush of fans we started talking about what the real purpose of the visit was. It wasn't just a project we wanted him to help with, we actually wanted to play him and we had a very strong schedule.

Oh, okay, just the program is very easy, so it will be easy to play. Many people thought it was decades away. Some people thought it never would be because they felt that to be successful at Go you needed human intuition. Go is the oldest board game in the world that is played continuously and, in a sense, is one of the simplest and also most abstract. There is only one type of part. There is only one type of movement. You simply place that piece on the board and then your goal is to create a linked group. of your stones that surrounds an empty territory and when you surround enemy stones you capture them and remove them from the board you earn points for the surrounding territory and at the end of the game the person with the most territory wins, it seems very simple, but then sit down to play and you realize You realize right away that, well, technically I know what I can do, but I have no idea what I should do.

It's incredibly difficult for computers to tackle because compared to, say, chess, the number of possible moves in a position is much larger in chess, it's about 20. In go it's about 200 and the number of setups possible on the board is greater than the number of atoms in the universe, so even if you took all the computers in the world and ran them for a while. Millions of years, that wouldn't be enough computing power to calculate all the possible variations. If you ask a great Go player why he performed a particular move, sometimes he will just tell you that it felt good, so we have to come up with some kind of clever method. algorithm to imitate what people do with their intuition, nice to meet you hello, so with the fans hooray, we agreed on a best of five match and we agreed that it would be filmed, you know, we will treat it as a serious match, game with alphago aja hwang pushes the Stone for alphago and of course I think I will win the game with alphago because it's just a program.

In the first game I made a mistake and I lost again in that first game. I think something clicked for him: this was no ordinary Go program. not only were we doing the same thing as everyone else, that something new was happening in the second game, I try to change my style, but the problem is that I also lose the second game and also in the third game, in the first game, even In the last game, I lose five. zero alpha go win the whole game after losing lose lose you can feel the pressure from him getting heavier and several times after the game he said he wanted to go out for some fresh air I told him oh I can go with you and chat with him .

I said no, I want to go myself, I feel something very strange, I lose this program and I don't understand myself anymore, I was worried if he would come back, you know, he seemed very depressed and he spent about an hour outside the office. I came back completely changed, I thought Maddie, this is the first time in history that the gold professional human player loses to the program, she told me yes, we know, but for you, I know, it's okay, I told her no. and yes, I'm not happy to lose. game, but I'm very happy to play it in the story.

Artificial intelligence researchers solved the game of Go a decade earlier than expected. The computer called alphago was able to beat the human European champion. Artificial intelligence researchers made a significant breakthrough. It really is a big leap forward, there is a big difference between the way the IBM computer beats Kasparov, which was programmed by expert chess players, and the way the computer more or less learned on its own, the way What we start to train alphago is by showing it a hundred thousand games that have been played by strong amateurs that we have downloaded from the Internet and first we make alphago imitate the human player and then, through self-play reinforcement learning, it plays against many different versions of itself. millions of times and learn from their mistakes.

The specific ideas that are driving alphago are going to drive our future the technologies at the heart of alphago are what are called deep neural networks which essentially mimic the network of neurons in the human brain it is a very old idea but recently due to increases in computing power these neural networks have become extremely powerful almost overnight large neural networks operating on big data can achieve amazing things alphago found a way to learn to play go learn is the key here is machine learning all the beauty of this type of algorithms is that, because they learn by themselves, they can go beyond what we, as programmers, know how to do and allow us to achieve new advances in areas of science and medicine, which is why alphago is an important step towards that final goal that my wife called me to. she told me don't watch the internet don't connect to the internet because people talk a lot about you with olivago fanue lives so long in europe he's not really professional he's just an amateur player this is very difficult for me so difficult The world was skeptical about what how strong alphago really was and how much further they needed to go to be the best professionals, so our program is improving over time and we want to push the AI ​​algorithm to the limit and see how far this type of self-control goes.

The improvement process can continue, so we needed to look for an even greater challenge. A fight like no other is about to begin in South Korea. Lee C Dole will be a long-reigning Alphaphage and the ultimate showdown between humans and machines. This is a great moment for both of us. the world of artificial intelligence and I think the world of alphago has overcome all the challenges we have thrown at it, but we won't know his true strength until we play someone who is on top of the world and who probably didn't choose him. Lisa Dole because she wanted a legendary historical player, someone who has been recognized as the best player of the last decade.

I won't be too arrogant, but I don't think it will be a very close match at the level of the player that Africa reached. con in October is not at the same level as me, so since it's only been a couple of months I don't think it's enough for him to catch up, my hope is that it's five zero for me or maybe four to one, so The critical point for me is to make sure I don't lose one more. We don't know how well our system will work against someone as creative as Lisa Doll. Furthermore, he is very famous for the very creative fighting game of him.

Yes, this could be difficult for us, but we will see that maybe in some ways he will be the most difficult opponent we can face. There were still a lot of doubts about whether or not they could beat someone like Lee Sedol von. player but he is nothing like the best players, he said that everything is a nine-down professional. Nine-gift pro is the highest rank you can achieve and follows the fanfare of the ranking system. Who is the European champion? He's just a two-gift professional. On the other hand, the same. above he said that everything is nine but professional they said that everything is going like roger federer to tennis he is playing at wimbledon to win the grand slam and it is not just this year he will be there next year and the year after and the year after that, of course, on the internet and in the whole gold community, everyone maybe not 100, maybe 99.99, I think the listener will win, it's very easy, it's taken seriously in Korea, it's as much a part of the culture as breathe or, you know, take swimming lessons or something on the surface.

It is a game, but inside it has a very deep philosophy. The board reflects the individual who is playing. The truth will be shown on the board. You won't be able to hide it in ancient China, Japan, Korea. Go is one of the four noble things like achievement. literacy with music, poetry and painting so that people think that go players are very intelligent and very noble. Master Guan started this school in order to produce a great prayer in Korea. He said that when he was eight years old he came here attending 9 a.m. classes. 9 p.m. seven days and then he stayed with the teacher one was already a little boy I remember when I was a student studying go he was very young and rural enough to think that pizza grew on trees touch things that are interesting wherever you feel like is beyond winning and losing, he wants to do something that is innovative or takes things to the next level, so every go player started his game with confidence, about 10 years he dominates the world of professional gold, he won 18 world championships, definitely , Lizard is going to winthe game. 5-0 yes, the whole game, he will win the whole game.

We had our evaluation match last week, we won a game and we lost a game and we lost the game in a way that would have made us look extremely foolish if that happened publicly, it means that we still have work to do and we have to take this very seriously. . It's too much of a risk that we could actually lose overall, not only that, but we could lose in a way that makes us look pretty dumb, so I guess yeah, I did it. If you want to say something about what you're trying, I'm walking home, so we're working 24 hours a day right now, training our algorithms even more trying to keep improving gradually until such time as we have the match that we have. brought together people with different skills joins the same team, so we have researchers, engineers, evaluation guys, you mean what happens if they create something new that was not used, then we don't see it well?

The first thing we should know is that things can help, the perfect solution is the leaf. programmer and built the original search engine so the management responsibility is quite big he will be the one sitting in front of lisa dole and actually playing the alphago moves makes me feel excited yeah at least I know He is a great player and I feel honored to be able to do it. playing with him many of my friends are very excited about the game and they keep telling me that everyone is watching just get ready alphago hello hello dennis can you hear us?

Yes I can hear you, can you hear me? Okay, yes, perfect. I was curious. meeting you, an amazing developer who created alphago, it's a pleasure to meet you, thank you, thank you in the same way, it's a real pleasure to meet you, I love the game of go, I'm not very good, but I don't know if you know that. um, but we actually have something in common. In fact, I also trained in a game when I was young. I used to be when I was a young professional chess player. I used to play for the England team and then I stopped playing when I was about to. 14.

I was um when I was 13 years old. He was the second highest rated player in the world after the match. I would like to propose that I give you a teaching game of Go and you give me a teaching game of chess. That sounds really good. It would be a great honor to do so. I'm not sure if it's okay to ask you this question, but I watched the games against Fanfuy and I didn't think he was at the level to play with me, but I heard he's getting on. really stronger, may I ask how much stronger did he get? um yeah, can't say much of course, but it definitely got significantly stronger with more time, I think a logical approach would be to follow Ash's local search and run it in fast alpha and generate a completely new data set and iterate over more time, but we ran out of time, we realized that if we wanted to prepare for a gigantic task like facing Lisa Dole, there would be nothing better than talking to a professional and we couldn't have chosen a better person than fanway, we invited him as an advisor because during the match we discovered that he is a man of good spirit welcome back thank you very much he is very valuable we had fun with us he was crushed by avago but then he became very positive and was a great help to us when they said to me, can you come back for help us make Africa stronger and feel this respect?

Of course, I accept this. I play with alphago to understand where the strong point is. alphago and where is the weakness baby i play this morning afternoon all the time and i found something i found a big weakness about alphago the great superheroes always have a hidden vulnerability right and the same goes for alphago he is incredibly superhuman in general but he has some weaknesses particulars, in some situations we may think that there is this space of all the things that you know and you know most of it extremely well, but then there will be these complicated clumps of knowledge that you just understand very poorly and it's very difficult for us to characterize when it's going to come in. in one of these lumps, but if you do, it can be completely delusional to think that you are alive on one part of the board when you are actually dead or vice versa, so there is a real risk that we could lose the match.

Everyone on the team will try to work harder and harder to solve the problem, but I think it is difficult to solve it very quickly. We have version 18 in the works. It is not like this? We will only use verses. 19 if it makes it significant, if it doesn't we had to rerun all the tests because unfortunately they went wrong so we're basically a day behind in our evaluation because of that, all the trust, well maybe let's just need to be realistic, we've tried a bunch of things and we've gotten to the point where we said we'd start freezing the code and saying "look, it's not" and actually not solving these delusions is still a realistic possibility. for the game.

We have some weaknesses that I don't think we're going to

full

y address before the game, so that's causing us a little bit of anxiety. The woolly doll is preparing for the fight on Wednesday live on the Internet. This South Korean professional goes. The player will face the artificial intelligence program Alphago tomorrow. Julian and George pack up version 18, save it to a laptop, and fly to Seoul. When I get to the hotel, I'll catch up with the team on what we need to do. the remaining time is to list the things that could go wrong in the soundness of the system in a difficult situation he likes to smoke yes there is a terrace and we will have security so he can go up and then be alone when he arrives Here I did not expect the attention at the game, It was literally front page news about eight million Koreans playing Go and even those who don't know recognize Lee Sedol, he is a national figure and that's right, there is something national. pride involved, but it's more than that, just the idea of ​​a machine playing against a human in something like this, I think it's inherently intriguing to people around the world, they gave him the pressure to listen before this, he played the tournament by country by itself but this time played for the human I really hope we win this first game, if you lose the first game you literally have to win three out of four yeah which is hard work where are you going to be?

I'm going to start in the match room and then I'm going to come in here to make sure everything is here. Yeah, you should probably be here. Hello and welcome to the Deepmind challenge game, round one, live from all four stations here in Seoul, Korea. I'm Chris Garjok from the American Go e-Journal. I'm here with michael redmond nine don professional welcome michael I want to congratulate all the people watching around the world well, the excitement is quite palpable here at the hotel I have never seen a crush of interest for reporters from all over the world world, okay friends, you are here, you will see the story done, stay with us for five minutes, guys, five minutes.

I thought we should take a moment together and think about what's about to happen. I'm very excited to be there. I am very proud of each one. of you and what we have done and we are not lost, I think it is just amazing that we are here, well foreign guys, oh thank you, we worked very hard to make sure this would turn out technically well, we tried and tested. I tried it and still there comes that moment where you're live, all the TV cameras are broadcasting everything and now it has to do what you built it to do.

I was a little nervous. It's the first time I sit down. facing a world-class co-player and I can really feel the spirit and courtesy of a great co-player like Lisa because I think it's the first time she's faced a strange opponent. I think he is not a human, he has no emotions, he is cold. but he remains very calm and I can feel his mental strength oh here is meow in kim uh nine don pro from the korean association baduk we are here live at the four seasons hotel on the 21st floor what are we watching how the game is going oh It's fighting, no, from the beginning, from the beginning, yes, he, I forgot the place very well, yes, just that he is like a great professional, like a great professional, yes, he is very aggressive now the fight is getting really complicated.

Actually, this is the first time. I've seen Alpha play a game that has this difficult overseas fight. I just saw him looking at his opponent's face and that's just kind of a habit, just an instinct as a player to look at the person in front of him, yeah, it's kind of like that. um, something that lisa would do when she was wondering how his opponent was feeling, it's just a habit so it's not like ajah huang was going to give anything away because ajahn isn't an alphage, oh he borders on human when you play. you can have an exchange feeling that I look in you I know, it's okay maybe you don't talk to me maybe you are afraid of me I can feel a lot of things but alphago you can't feel anything so when you don't confirm anything when you play you have more than more questions about yourself to begin with you think it's okay my movement is good and very good very good oh maybe bad oh terrible why would you listen to more and more?

It's hard to know where to be. All the different rooms are exciting in a different way. It's not to be here in the heart of the operation yes, I feel safe here. White is thinking of making this big invasion here from his thick wall. Yes, go ahead, he's already there. He is not a safe face. He is quite horrified by the fact that he can. I can't believe what I see now oh she really thought you know she's a little behind and she made a very aggressive move here okay make it very complicated okay I think she said oh if she doesn't know answer quickly then it can collapse.

I want to say that this is something unthinkable as a human group. What if she knows everything about what's going to happen next? Where we are? Here are our search terms. 50 or 60, that's the maximum number of moves ahead that alphago is looking at from the current game position. It's usually over 50, often it's over 60. In games we often see around 150 alpha going to kill, now we're 100, move 115, so we're getting to that point, we're all just amazed in the middle of the game because Although Go seems to be doing a much better job than we all thought, I thought they said everyone would lead the game comfortably, but it turns out he's struggling right now, but I think he'll eventually prevail, I hope I did. . uh you know more than five to zero, but now I'm not sure there can be the first error, a clear type of error, strange, why one or not right now, mark this if it's like this white has won by far, yeah, wow.

It's so shocking. I expected Alphago to win only one game. I can feel his pain as if it were him. I could not believe it. You know I couldn't accept it. It takes time for you to accept the result. Maybe Alphago is very strong now, but no. I will not believe the strict rules, it is just the portion of goal that we take from the player, we cannot believe this because the form for us is something very far away, it cannot come now, it is impossible, but in reality it is not, I think he resigned in a very polite manner these are he is black but he blew white stone I think he resigned oh my God oh yes, time has stagnated wow, I feel very good, I feel that I really believe in alphago, of course, it is natural that humans want humans to win I mean, I think it's a natural response, but alphago is created by humans and I think it's the ultimate sign of human ingenuity and intelligence.

Everything alphago does it does because a human has created the data that it learns from and created the learning algorithm that it learns from. that data creates the search algorithm, all these things come from humans, so really this is a human effort in the battle between man and machine, a computer just got the victory, the deep mind put its program to the test. computer against one of the brightest. minds in the world and an alphago defeated a professional player who has 18 world go championships under his belt, the winner is considered a breakthrough in artificial intelligence in research, we normally work to produce an academic article, it is published and perhaps we can talk about it. at a conference, if we're lucky, this is not normal for research.

In fact, I've never experienced media attention remotely close to this, so it's a special moment for all of us and we're just enjoying it while it lasts in a showdown between Man and Machine are the ones who have won well so far. . It's the machine from Seoul, South Korea. The computer with artificial intelligence defeated the world champion in the ancient Chinese board game. Go Lee Sidol lost the first matchup, but has four more chances. It's a little strange being in front. cover and all, as a computer scientist you usually sit in your corner, it is possible that no one really knows, maybe you have heard the joke, how can you tell that a computer scientist is an extrovert and not an introvert?

If he's outgoing, he looks at your shoes when he's talking to you instead of just, like you look at Archer, he avoids all the cameras like crazy, it's like the game is over and he's back and back in his room and I think that many scientistscomputer scientists would be like that. we were more about doing our job than being in the spotlight hello and welcome to game two, round two of the google deepmind challenge, showdown between man and machine, game one, machine defeats man, high impact headlines in everyone, the reactions on the ground here from people in Korea were just stunned, they estimate that 60 million people watch the game in China alone, which probably makes 80 million people watch this game worldwide, it was just unbelievable and if anything it's probably an even worse house today so he said, He'll know today, he knows this is a really important game, right, he's got to win.

I think he's probably looking for a little revenge. My impression is that he perhaps underestimated Alphago and is going to change his tactics. Well, if I say. anything about alphago that isn't normal, maybe it's the way he handles a game when he thinks he's ahead. Yes, actually we are going to have a visit from one of the team members and we will talk exactly about that point from the inside, thank you. Much to come to Thor, I really appreciate it, can you share a little about what's going on at alphago? So alphago has these three main components, there is the policy network that was trained in high level games to imitate those players and then we have a second component we call it net value and it can evaluate the position of the board and say what is the probability of winning in this particular position and then the third component is the tree search where you would look at different variations of the game and try to figure out what will happen in the future, so if we now take a position This way, first the policy network scans the position and finds what would be the interesting place to play and builds a tree of variations and then employs this value network that tells you how promising the outcome of this particular variation is, so alphago tries to maximize your probability of winning, but he doesn't care at all about the margin by which he wins so when you see slow movement maybe it's an indication that alphago thinks he has a good chance of winning yeah that's a little giveaway we're looking for a foreign hotel ooh it looks like it looks like lee is taking a little break lisa dal go smoke and alva go just play don't sing about the opening it will be zero or not so i jab go to alphago make move 37 and aja puts the stone in the board that's a very surprising move I thought I thought it was a mistake when I saw this move to me it's just the big shock that humans normally I will never touch this one because it's bad it's just bad we don't know why it's bad but it's kinda stop, yes, it's the fifth line.

Normally you don't do a shoulder here on the fifth line, so it comes on top of a fourth. the line zone is really unusual, yes it is an exciting move. I think we've seen an original move here, that's the type of move that plays, go for interesting things, this fifth line shoulder head is like that. Yeah, I didn't expect that. I don't really know if it's a good or bad move at this point, the pro commentators almost unanimously said that not a single human player would have chosen move 37. So I took a look at alphago to see what he and alphago thought. actually agreed with that assessment, alphago said there was a 1 in 10,000 chance that move 37 had been made by a human player, so he knew this was an extremely unlikely move, it was beyond his human guide and came up with something new and Creative and different.

I'm watching the game a lot through these commentators. That's how it works, so when they're confused, I'm certainly confused at the same time. I hold on to the fact that they are confused. an interesting moment when everyone else is confused, who is not confused right next to the machine, i want to see lisa dal when i see this, she is back, lee is back, foreigner, i have usually seen one, two minutes, no more, but this time I think it was more than two minutes. the more I see this movement I feel like something changed maybe for humans we think it's bad but for alphago why not go it's like geopolitics how something small that happens here can have a ripple effect hours later down the road in a different part of the board of the game like it rotated on its axis at that moment this move is very special because with this move all the stones played before work together they connect looks like a network link everywhere it is very special very special points this is a difficult game for lisa alphago just won't let Lisa do what she wants, right, the black hat is almost 50 points, that's a lot, not a good sign, oh, he said, all he just slapped on the side of his head, oh, wow, I think black is ahead at this point, it looks good and we.

You are not a stable and stable path now I only see it lucidly, it loses a lot. We normally resent people a long time ago, but they won't try, they played, they played, they played, they just don't want to reason because then, white, oh, they quit, it seems. lisa dole just resigned there was a great sadness all over the floor you could feel it during the game i felt it during the game and i left the commentary room to go to the press conference and someone stopped me another tech reporter now . At first, all he wanted to talk about was technology and how cool he was, but then even he fell into that moment of melancholy where he was upset too.

I am quite speechless, I admit that it was a very clear loss on my part. From the beginning of the game it wasn't a moment where I felt like I was leading the game, you feel elated and a little scared, there's something that I think scares people about a machine that learns on its own. we, alphago, is obviously just a computer program, but looking at the comments on the Internet, I already saw that commentators called alphago as he and she during games completely unconsciously, and alphago is really a very, very simple program , it's not even close to

full

AI and we.

I already see that happening, so I find it very interesting that the tendency to anthropomorphize AI systems is one of the big obstacles in the way of trying to understand how AI could affect the world in the future, for example, the conversation is about what could go wrong. What are the risks? So you invariably see this Terminator image whenever there are these glowing red eyes. True, we are closer to a smart washing machine than to the Terminator. If you look at AI today, we are really very nascent. I am extremely. I am excited and passionate about the potential of AI, but its power is still very limited.

I think people are right in thinking that there is a danger that as we continue to improve these systems, we could lose the threshold that we cross. danger, but the good news is that there are already people thinking about those dangers, you know, there is a lot of talk now that we are leading the discussions on this that maybe there should be some kind of cross-industry working group on best practices or something where the leaders of the research teams in those organizations, you know the big guys who are working on ai ibm microsoft etc., get together and ensure that ai is used ethically and responsibly.

I think the important thing is that there is this community of people who are leading. uh the cutting edge of AI engaging with academics and already thinking about the long term and how we can ensure that innovation is responsible as the power of these machines increases even further. That's it, folks, day three, game three, let it all out. Back to the wall, he is down 2-0, he has to win today to keep hope alive to console him and review Africa, console him. Do you think he's upset or I mean, is he upset? Yes, at the beginning there was a terrible fight and then Avago played very well, so they secured a very early lead from mu 50.

The win rate was already very high and was climbing towards 100 percent, oh, that's a big one. movement, what is your probability? um uh, the blue and white shape that he tried to fight. directly into the game but it's not even half obsolete when we change our style to play with the opponent it's usually very very bad so the game is easier for alphago it looks good for us locally I think the black group is huge and has nowhere to go. and it's going to be going around I don't know how to describe the situation if I were black I would quit we should admit that we are facing the strongest existence ever seen in the history of gold there is no point in playing the end game and you're going to know that you're going to lose as well That even if the black can live there, oh, he quit, okay, wow, wow, you saw history made here tonight.

Alfago is one again three wins in a row three wins in a row he has won the match when lisa is in the game he, he seemed unhappy, it's not just that he lost the torment, it's especially about this game because he doesn't play his game, I'm very tough with this, but I can't do anything, you know, certainly, I feel a little ambivalent about it. Since I am a game player and you know that Go is the pinnacle of board games, but I really like this statement that one of the top Chinese professionals said. You know, if Alpha Go wins, maybe we'll really start to see what this game is.

I couldn't celebrate, it was great that we won, but a big part of me saw this man trying so hard and being so disappointed. I see the Internet, a lot of people talk about the listener, maybe I don't. play the best you know, we're players, okay, sometime in China, in Korea, Japan, we see go like ah, we're artists, you know, we play our best, so please be nice to Lisa Doll, he He's very, very good players, excellent players, I'm in In the room I see this door, he won't win, I try everything, we just can't do justice abroad.

You could see that he was more relaxed after losing three games in a row and he was saying that the stakes were high, you know? in the end it's not about pride they don't feel confident but they feel light can lee saddam find alphago's weakness is there in fact a weakness? I was thinking I just pulled the plug you just pull the plot everyone keeps applauding for this at all and then yeah it's not going very well now I'm worried it feels pretty good for black at this point it feels pretty good for black. I want Eliza to play her game because at the moment she tried many things to play with alphago to understand. alphago but he never tried to play himself i thought the man kept alphago looks like the real mirror when you play with alphago you feel very strange you look like you and everything becomes naked the first time you see this you want to see it because oh , it's me real me and more you need to accept oh, I'm a real me, so how can I do it now?

It's turning into a very, very dangerous fight. This is really Lisado's kind of game. He likes this type of fight. White has to find something within black's territory. I think he's already planning to try something, so you believe in Ethereum's ability to live in a very small area, right, yeah, there's a little bit of potential there and I think maybe he'll try to do something, he said, all magical, oh, what? What would be the magic move Lisa is running out of time but she will have to use all this time there it only burned like seven or eight minutes just in this move I don't see anything anymore we can't find anything What is Lisa's doll?

So far, yes, he is very concentrated, he really is, look, look, Lisa is very patient, wait, wait, this moment I feel something that feels like the wheat of the wolf in the forest in the winter, he is cold, I feel a lot, very cold, but I need patience. but the moment he arrives they come out to attack this is the hinge of the game oh look at that move it's an exciting move no basically oh he found the wedge it's going to change the equation because now black can't escape that would be great if that works alphago ends to play something maybe unusual you know, I'm actually not sure what alphago is trying to do here what it's about uh you don't really understand it well that was a sharp drop in this could be that it actually can't find a way to get past I think this is far enough away that I can see that it doesn't work and now maybe it's tilted I don't know, it looks like it's like a viewfinder Yeah, you made a mistake, did something strange happen there? no, it will look normal, well, here we go, well, we can definitely say, well, there is a weakness, we would definitely say the mistake.

I felt this mixture of this sinking feeling in my stomach where I was wondering if alphagus was becoming delusional in this situation where I could see that he was starting to play strangely and at the same time I was relieved by the lucidity that he actually had a chance. Now I just like to try not to look horrified. I knew after moving 17, after 10 or 20 moves, I saw strange moves from Avago. And I know Abago somehow went crazy, but I didn't realize why when I was playing we went for 90. At that time, okay, I think something went wrong, that's the longest investigation in the entire gaming business , yeah, I think it's like your search, so I do it.

It feels like alphago has gone off on a tangent, what are you doing right?Maybe you have a master plan, no, you don't even think you have it, you do it, then you know it's done because you start, you start to evaluate it in a different way. look look at lisa she's confused I was like what is she doing that's not uh I'm scared confused that's uh what is she doing why is she going you know you asked if it was a mistake I've said before yes yes yes, of course, yes deepmind has discovered how to write code that doesn't have bugs, that's bigger news than alphago oh you're kidding hmm literally let's play the next

movie

i think i think son's gonna laugh maybe he's gonna laugh oh not really I know what alphago is trying to do here that's the euphemism of the year it's love no that's the movement hong doesn't make mistakes he's very confused these are not human movements this movement too um kind of inexplicable I mean those you can you can clearly call those mistakes, yes of course, but it's the first time in the four games that we've seen moves like that, right, oh, the value fell even more, it's strange, we're like 45 lights. and white, white, swing, come on, at least this is amazing.

Okay, he was so sure that this black moya would consolidate with the black and there's nothing there and he somehow erased everything like he disappeared, so he found the weakness of him. The wedge move probably surprised him yeah right yeah alphago seemed to have such a good game and then this whole sequence in the center uh kind of changed and he pulls off a miracle. He manages to make it so complicated that the artificial intelligence doesn't evaluate it. that's right, i think it looks like aj huang maybe sitting there also knows the game is over. I can't, I can't, I can't stop smiling.

He smiled? No, I haven't seen him smile yet. I think he's still finding out that he is so. Seriously, he wants to be careful. Yeah, is he very worried? she really does her best, I think he feels something like, you know, excessive responsibility or burden, it seems like alphago is resigned, wow, the most incredible game, I'm almost going to cry, it was the fourth game where he comes back and wins well, abroad, I heard a lot of people say: you know, they were running in the street, they were so happy, you know, they were singing, they are celebrating, especially after 0-3 down in that moment, it seems hopeless that the end of the world is approaching, but we see the light.

Usually I'm happy when I win, not how my colleague wins, but this time it felt like it was my window. My question is about the 78th move, the Chinese cue player Kuli said it was a work of God, what? Were you thinking when you made that foreigner? It leaves me a little in awe of the power of the human brain, particularly Lee's amazing ability to cause alpha trouble and find something seemingly out of nowhere, so we really want to understand what happened and what it was really like. Lee's move that the key is the center. If the center continues, there are notions that we were winning before, but if alphago thought he was winning and couldn't convert that victory, that means he wasn't actually winning, otherwise he would have had the right answers.

We've played it, there's a robbery there, what probability does 0.007 give you? It's just that the position he played in was when he played well, so we thought this was the one. Yes, the value is very low, so the God move was literally a God. move because we think only one in ten thousand uh humans would have figured out that that's right, it doesn't occur to me, it's the first five of the first five moves that you would consider and I said, Lisa Doll, he said that was the only thing. and thought he was the only one, yeah, welcome back to season four, world champion Lisa Doll went looking for Alphago's weakness in the fourth game and found it today in the last round, she's looking to see if she can repeat that, we'll see what's happening.

The place is a madhouse downstairs, maybe as many, if not more, journalists here today than there were at the first game. We were all very excited for the fifth game to see what would happen, if this was going to be a 3-2 or was it going to be. it's going to be a 4-1 thing and you know there's a different message with both for me, I don't think alphago will make the mistake again but who knows, maybe he'll tell me and the licit ones are more confident in this aspect now. maybe lizzita will find a key to open the alphagus.

You weren't crazy at the time of this move, yeah, yeah, and now you don't approve of this, yeah, I'm thinking maybe alphagus hasn't recovered from game 4 yet, yeah. so white has to do something with this later, i don't expect he doesn't know he needs to play, although making any move is bad, i feel something, oh maybe this victim has come back, are we seeing another short circuit or is there something that is going? uh I think it might be kind of a misinterpretation uh and we're pretty comfortable saying that you're doing well for at least well for at least no, there's no reason for White to play that move, it's bad to move and in some cases too he's going to lose a point it's exactly the same 91 certainly someone was telling me that maybe throughout the whole game we thought alphago was wrong with the board position we were very worried that oh he was going to play rubbish he's going to be loose in a way very embarrassing and this continues throughout the game, but this is a very strange moment, when did they play?

Yes, it turned out that none of us knew how to go well enough to accurately judge what alphago is doing. Because? it's sweden now this one looks like white is winning ah we all say that some of our pay moves are very strange and strange and baby mistakes, but after finishing a game we have to doubt ourselves of our judgment alphago making another kind of meaningless throw-in that I'm not really sure what it is, this is 10 or maybe 11. The game seems weird and we don't fully understand it. I think it is important to study more about hypogo errors such as movements, and then maybe we can adjust our knowledge. about go to me the most surprising thing that came out of my understanding of go as a result of watching the alphago game is the infamous slack moves, well there is something strange about the way it is played because some moves are executed that don't They are really necessary, right?

Lazy movement is a movement that appears lazy. You can see these other better moves and Alvago is rejecting them, but what I think alfafago is teaching us is that we have been using the score as an indicator of the chances of winning, so the greater my margin of territory. the more confident I am that I'm going to win and I say no, no, it shouldn't matter how much you win, you only need to win by one point, why should I take over all this extra territory when I don't need it? The lessons Alphago is teaching us are going to influence how Go is played for the next thousand years.

Am I telling why I won this game? How much are we talking about? Colon is just a game, but we can learn important lessons from our If computers were so successful at playing Go games, the machines would have the ability to not only process a large amount of data, but also to analyze it intelligently, such as As in the case of Go games, the machines would make moves that surprised even the experts, and eventually the machines would win. our confidence because we will see that very often they make a better guess than we could have made as humans.

I think the whites might have a slight advantage here. What I would say is that it is the most surprising thing, yes, that I have ever experienced. You know, for us, it's the culmination of a 20-year dream. It started as a pure research effort. We just wanted to understand if neural networks can play the back-and-forth game. There it got to a level that I never expected and I'm incredibly proud of the team when I started doing AI, it was four or five years ago and I was really interested in it, but a lot of people would discourage me and let's say there's no future, so at the same time Seeing that in five years we are at this stage, it is amazing that there are so many possible application domains where creativity in a different dimension than what humans could do could be immensely valuable to us and me.

I'd love to have more of those moments where we look back and say yes, that was like in movement 37 something beautiful had happened there, at least in a broad sense. The 37th movement begat the 78th movement begat a new attitude, at least a new way of seeing the world. game that improved through this machine your humanity expanded after playing this inanimate creation and the hope is that that machine and particularly the technologies behind it can have the same effect on all of us. I remember listening to a talk by Kasparov who said that a good human plus a machine is the best combination this is a unique experience no one can have this experience playing alpha phage fighting games like this so I hope that Lysador can find something in this game five maybe some change in his game that I see today fight, fight, fight, it's very good, grandmaster, true grandmaster, although I'm done, but despite the whole story, maybe it's just the beginning, we don't know , it's just when I play with alphago, he showed me something.

I feel beautiful, only then I see the world differently before. everything begins what is real inside the game go with this thing i will change something with my game maybe he can show humans something we never discovered maybe it will be beautiful for you