Soul of Chess: Buy the chessboard from Shi Guangjia at the beginning of the game

Chapter 256 Defeating Magic with Magic in Round 255

Kuwahara Jiudan muttered in astonishment all the time, "It shouldn't be, it doesn't conform to chess theory."

Yu Liang, who always thought he knew Tian Minze very well, fell silent, and he couldn't explain why.

Because from the perspective of the situation, Tian Min seemed to give up, throwing pieces on the chessboard at will.

On the contrary, AlphaGo is indeed a very well-organized start. Xing Xiaomu and Xiao Fei guarding the corner are still common routines of AI Go.

Although no one can understand Tian Minze's Go, the commentator still dare not comment lightly. After all, the person sitting there is No.1 in the world today, and many people can't understand his Go.

But Gao Yu still expressed her own opinion on the fourth dan, "As an average chess player, I think Tian Minze is a little bit off today."

Kuang Yuan and Yu Liang smiled at the same time.

It has to be someone's professional commentator who can speak.

Behind this unstructured way of playing is some logic of Tian Minze, which is obviously the frequent escape.

But it is an exaggeration to say that it is off first. It is obvious whether it is running to the corner to play chess, which is incomprehensible, but from the situation, it seems that it is difficult to say that Tian Minze has no chance at all.

Because AI doesn't seem to keep up with Tian Minze's rhythm.

In short, Tian Minze's strategy is as follows:

Slowly connect a series of chess pieces on the outer circle of the chessboard, trying to surround this series of chess pieces into a large "encirclement circle". At the same time, lay some scattered chess pieces in other corners of the chessboard to distract AlphaGo.

AlphaGo has been obsessing over the pros and cons of some chess shapes, but it doesn't seem to notice that Tian Min has quietly surrounded a large area.

Tian Minze's field is not far behind, and the limited number of chess pieces that attract AlphaGo's attention are also playing a role. They all choose the kind of placement that is difficult to be eaten by the opponent at once.

"Did Tian Min find the bug of AlphaGo?" Yu Liang guessed.

Kuwabara's nine-dan is naturally not so keen, but he just thinks the whole game is very strange, and he can't tell what is strange, "What kind of BUG?"

Yu Liang introduced, "Tian Minze's chess has a very low winning rate, so AlphaGo plays the position with the highest winning rate, which ignores Tian Minze's small moves on the side and corners. This kind of play is against humans. Chess players are useless, but it seems the AI ​​has fallen for it."

Having said that, several people seem to understand.

Yu Liang's introduction was very simple and easy to understand, and even Internet giants who didn't know much about Go began to talk about it.

Tian Minze's way of playing goes beyond the scope of Go.The purpose of this method is too obvious, that is, to enclose a large area with chess pieces.

For human chess players, as long as the opponent has mastered the basic Go rules, there is basically no winning rate in this method.Because as long as your "encirclement" has not closed, the opponent can easily break the entire encirclement within a few moves.At this point, it is no longer possible to win the game because you wasted too many pieces on the encirclement.

But as far as the AI ​​is concerned, Tian Min's chess pieces are not efficient, so the AI ​​doesn't pay attention to them at all, and is still looking for a position with higher chess efficiency to play.

Although Tian Min adopted an unconventional way of playing chess, he was still very careful, hesitating every move. On the contrary, the AI ​​played very quickly, almost without thinking, as if the chess game in front of him was not difficult for it.

However, Dr. Huang, who was sitting in front of the chessboard, had an unusually ugly face. He already clearly felt that a bug in the computer program had been discovered.

In the backstage preparation room, AlphaGo's team was equally unbelievable. The winning rate given by the AI ​​had reached 90% for one's own side, but it was obvious that there were more and more spaces surrounded by black pieces on the chessboard.

Tian Min got up and went to the window to get a glass of water, and calmed down for a while.

On the contrary, the chess fans and all the staff in the spectator room are already boiling, because just after the next move, as long as the black piece is dropped, three pieces of chess around the goal have been completed on the board.

AlphaGo turned a blind eye to black's moves, could not recognize any potential danger, and did not destroy black's "encirclement" before it closed, allowing black to play.

There is no way to continue the on-site commentary. After all, this has gone beyond the scope of Go, but the Chinese Chess Academy is also prepared, and will soon invite an AI engineer from Tomorrow Information.

Facing the camera, the young engineer was still a little nervous, but he quickly talked about his profession and became eloquent, "Because AlphaGo was developed based on confrontational drills. In its confrontational drills, it did not experience After enough of these moves, he's short of samples.

In other words, Tian Min’s creation of “encirclement circles” is doomed to failure in the judgment of AI, so AlphaGo did not realize that if this form continues, he will lose.

Without the ability to judge as a basis, the AlphaGo program can no longer operate. "

As the engineers expressed their own judgments and interpretations, everyone gradually understood that this kind of play has never been seen by AI, so naturally it cannot fight against it.

Gao Yu sighed in the fourth paragraph, "Tian Min is using magic to defeat magic."

With the 207th move, the black chess fell, and the AlphaGo team in the background saw that the program entered a long period of thinking. After 20 minutes, the winning rate dropped from 90% to 40%.

Afterwards, the computer started to make old mistakes, and played a chess game that he thought was a general.

But Tian Min ignored it at all, and continued to close. After another 10 or so moves, AlphaGo calculated that his winning percentage had dropped to 0%. burden.

Tian Min stood up again, shook hands with the shocked Dr. Huang, and walked out of the game room.

All along the way were media of all sizes, including news channels, sports channels, and entertainment channels. All kinds of reporters were on both sides of the road. When they saw Tian Minze passing by, people began to applaud involuntarily.

Soon there was thunder in the palm of the hand, and everyone had some excitement and admiration.

For the first time in the man-machine battle, it ended with a human victory. Although this victory was a bit sudden and a bit opportunistic, it was a victory after all, and it was a key victory that human chess players needed.

Tian Min walked into the post-match interview booth, and it was still a routine interview at the end of each game.But this time, all the reporters were hurriedly deleting their articles.

Most of what everyone prepared before was the press release after Tian Minze's loss, and it was going to be replaced by Tian Minze's victory press release, which caught everyone by surprise.

Tian Min didn't feel much joy, and remained calm as usual.

The reporter's first question was, "Tian Kudan, may I ask if you have found the Achilles' heel of AlphaGo? Will you consider using the same method in the next game?"

(End of this chapter)