AI AlphaGo from Deep Mind defeated the world champion in the logical game of Go

In the sphere of the development of artificial intelligence, a very significant event occurred. The AlphaGo program, developed by Google's DeepMind division, defeated the world champion in the logical board game, Korean Lee Si Dong, as part of the first of five historical games to be held in Seoul. Lee lost the first game after three and a half hours of play, while on the watch there were still 28 minutes and 28 seconds left.

This is the first competition in the history of this game between a professional player in the 9th, the highest dan and the computer. The main prize of the competition is 1 million dollars. Who will ultimately take the final victory, we will be able to find out in a few days.

"I was extremely surprised. I did not expect a loss. I did not think that AlphaGo would play in such a perfect manner, "- said the world champion in the game go.

The founder of DeepMind, Damis Hassabis, expressed his "deep respect for Lee Ci Dol and his incredible skills," also calling the game go "incredibly exciting" and "very tense". AlphaGo team captain David Silver noted "the amazing complexity of an interesting game of go, which made their brainchild AlphaGo work almost at the maximum of its capabilities."

For those who do not know, go is a logical board game with a very deep strategic content, originated in ancient China according to various estimates 2.5-5 thousand years ago. It is considered as one of the most serious tests for AI, as it contains a huge number of possible combinations of moves. Despite the fact that computer programs have already pretty well mastered such games as chess and checkers, and even won victories over world champions in these disciplines, the high level of intuition and strategic thinking that is required of a player in go makes it much more difficult than the above games and very difficult to learn for a computer.

DeepMind's AlphaGo program is currently the most advanced among all the others and is built using a very complex and multilevel neural network system, and also has the ability of machine learning. Thanks to this, AlphaGo was able to defeat the European champion in the game of Go Fan Hui last year. Therefore, a kind of new level for her is a Korean player and world champion of the game, Mr. Li Si Dol.

"I do not regret accepting this challenge," says Lee.

The Lost Champion

"But I'm shocked, I really admit it. But what's done is done. I really enjoyed playing with the computer, and I'm really looking forward to the next game. I think I lost because I literally passed the start of the game. So next time I will pay more attention to the beginning of the game. I think this will increase my chances of winning. "

Lee notes that he was deeply surprised at how powerful AlphaGo had started the round and what unexpected moves for him had been made.

The next batch of Lee and AlphaGo will be held tomorrow. The remaining three will be held on Saturday, Sunday and Tuesday. No matter how this series of games ended, today's victory of AlphaGo is a huge step in the development of AI. However, Lee is eager to take revenge.

"Today's game was very tight and tense," says Hassabis.

"When the time came for the beginning of the first game, we really did not know what it would all turn out. And now we do not know. After all, there are four more games ahead. "

The article is based on materials https://hi-news.ru/technology/ii-alphago-ot-deep-mind-obygral-chempiona-mira-po-logicheskoj-igre-go.html.