Go: The live broadcast queen launched a gorgeous rebellion against the dog

Chapter 142 Puppy's Tail

Hearing this question, Li Jingzhe was stunned:
"Uh, Grandpa Huang, I don't know much about Go, so it's hard for me to judge."

"It's okay, just tell me your first impression!" Huang Baoyang replied.

Seeing that the other party insisted, Li Jingzhe prepared to answer.

Originally, he thought that in this era of rapid development of information technology, once some traditional human skills were discovered and surpassed by computers, there would be no possibility of them being surpassed again. So he subconsciously thought that OMEGA GO must be stronger.

But when the words came to his lips, Li Jingzhe suddenly felt something was wrong.

Since the other party has asked such a question, the answer should never be taken for granted.

Considering that OMEGA GO lost one game in the five games with Choi Sung-seok, the opponent made a so-called "God's move" (of course, as a quasi-industry reserve, Li Jingzhe also knew that it was just a software bug). With the further development of Go AI technology, human players can theoretically "learn from the barbarians to defeat them"...

Hiss...No way?

Li Jingzhe frowned slightly with doubt in his heart and asked tentatively:

"Is this Lei Qi stronger than OMEGA GO?"

Huang Baoyang smiled even more happily: "Jingzhe, it seems that you are not only smart, but also clever.

"Actually... this issue is very controversial. But I have to say that this issue is debatable and not outrageous."

Not outrageous...

Although Huang Baoyang did not give a positive answer, his expression changed when he heard these four words.

Huang Baoyang continued:
“In fact, this issue would have been largely uncontroversial a year ago.

"'How can humans beat computers at Go?' This is the common conclusion of the people, just like countless people in the past firmly believed that no matter how powerful a computer is, it is impossible for it to beat humans at Go."

Hearing this, Li Jingzhe smelled something.

"You mean... in the past year, the level of human Go has improved a lot, right?"

"Yes." Huang Baoyang nodded: "The top Go players improve very quickly.

“A year ago, most people believed that although humans could gain a certain advantage over OMEGA GO in the layout stage by learning the most advanced AI formulas,
“But once the game enters the middle stage, as the situation becomes more complicated, these advantages in the layout stage will be quickly recovered and surpassed by OMEGA GO with its powerful computing advantage. Considering that humans are far less stable than machines, humans are ultimately no match for AI—even if it is only the first generation of AI.

"But now... whether OMEGA GO can steadily recover its disadvantage in the mid-game is a big question mark."

Having said that, Huang Baoyang handed over his cell phone.

All I saw was the current game of Chess Saint War on his phone, but with all kinds of curves on it.

Li Jingzhe took the phone and asked, "Grandpa Huang, this is...?"

"The AI ​​compliance rate chart of the two chess players. Now the VVVIP of Yiguo has this service. You can take a look. Ah, you should understand the terms such as first choice and first three choices, right?"

"got it."

Li Jingzhe, who works in computer science, is well aware of this common sense regarding AI.

But he was still surprised when he saw this data chart.

As the game was almost over, White's first-pick rate was as high as 70%, and the first three-pick rate was an astonishing 94%!

This is the overall AI compliance rate! The first three choices can reach 94%?

Li Jingzhe immediately looked up at Huang Baoyang and was about to speak, but Huang Baoyang spoke first.

"The detailed data analysis in the lower right corner also includes the specific layout, mid-game, and endgame...ah, it's the matching rate analysis in the final stage."

Li Jingzhe hurried over to check, but the specific data was even more shocking.

He originally thought that the first-pick rate was so high because the player "memorized the game" and made a choice at each step during the layout phase, thus raising the overall data.

But he never expected that although the rate of the first three selections in the layout phase of this game was as high as 98%, the rate of the first selection was only 63%!

On the contrary, in the middle game, although the selection rate of the first three dropped to 93%, the selection rate of the first one rose to 67%!

The finale was even more outrageous, with a 98% selection rate for the first three moves and a 76% first-pick rate! "Grandpa Huang, why... why is the first-pick rate of human chess players higher in the middle game than in the layout?" Li Jingzhe's voice trembled with surprise.

"Actually, this is not necessarily the case. Most of the time, the layout should be slightly higher."

Huang Baoyang explained with a smile:
“However, because there are actually many options in the layout, and the first three options are very close, as human chess players, we often do not act completely according to the AI’s choice.

"On the contrary, in the middle game, because the battle gradually unfolds, the chess player must try to grasp the most important point at the moment, so sometimes in some chess games, the first choice rate in the middle game is higher than that in the layout."

"But... the top three pick rate in the middle game is 93%, and the first pick rate is 67%?" Li Jingzhe was still in doubt: "Is this really something a human can do?"

Although he doesn't know how to play Go, he has a Go fan at home who has influenced him. He knows that Go is different from chess and Chinese chess. The mid-game and battle situations are complicated and there are many options.

As a result, under such circumstances, humans actually had a 93% top three selection rate and a 67% first selection rate? !
You know, with the development of AI in recent years, if the current Go AI were to play against the first generation OMEGA GO, it would definitely be a ruthless massacre!

If a person can achieve such a high matching rate with modern AI even in the middle game, it might be really difficult to recover the disadvantage in the layout in the middle game with the first version of OMEGA GO!

"How is it? Are our professional chess players working hard?"

Just when Li Jingzhe had not yet recovered from his surprise, Huang Baoyang said with some pride.

"Ah, of course, this was a game in which Xiao Lei played quite well, but it was by no means accidental.

“Like the top chess player on the stage just said, in the last round of the Nongshim Cup, South Korea’s Liu Shixian also performed extremely well, and the situation today was almost exactly the same.

"It was after that game that some people on the Go forum began to wonder whether top human players could compete with the original OMEGA GO."

"But...how could someone do this?" Li Jingzhe still found it hard to believe.

As a gifted person who aspires to engage in front-line computer science research, Li Jingzhe naturally knows that modern Go AI represented by OMEGA GO all use deep learning to reasonably reduce redundant calculations of chess games to ensure effective calculations with powerful computing capabilities that are absolutely unattainable by humans.

But now, humans can actually compete with modern deep learning AI? !

You know, before the emergence of modern AI, the strongest chess computer software “Snowfish”, its complete form has almost never been defeated by humans!

However, after learning chess for 4 hours using the OMEGA ZERO software derived from OMEGA GO, "Snowfish" was severely beaten by the opponent with a record of 28 wins, 72 draws and no defeat.

Among them, OMEGA ZERO's first record was 25 wins and 25 draws. This winning rate is simply shocking in the Chinese chess game where it is generally believed that the final solution is a draw!
Therefore, it can be basically concluded that in chess games, humans are crushed by deep learning AI.

But seven years after Go players were ruthlessly crushed by deep learning AI, they are actually starting to touch the deep learning AI that crushed them in the first place? !

After fully understanding this fact, Li Jingzhe was a little dazed for a moment.

The reason why OMEGA Software chose Go as its first deep learning AI work was largely to catch people's attention - what could be more eye-catching than a guide to "the world's most difficult board game" and "the last dignity of human wisdom"?

However, after the OMEGA GO software easily shattered the pride of countless Go players for thousands of years, it did not stay here for long.

After all, we are a business team. How much money can you make from a niche board game like Go?
Since then, top commercial AI has basically paid little attention to this niche project whose popularity is limited to three East Asian countries, and the most cutting-edge AI researchers are too lazy to pay attention to the situation in the Go circle after they have messed it up.

Only some AI companies in China, South Korea and Japan are developing and iterating AI according to the needs of the Go community, but this type of development is basically limited to iterative optimization, and there has never been any qualitative change in the core.

But who would have thought that after Go completely faded out of the cutting-edge research of AI, a shocking new situation would emerge!

Facing Li Jingzhe's question, Huang Baoyang raised his eyebrows:
"Jingzhe, an old antique like me really doesn't know this!

"But... maybe this is why we still love Go as much as ever?"

Just as Huang Baoyang finished saying this, there was a sudden new noise on the stage.

After the 286th move, Lin Ruixin, the black player, resigned and admitted defeat!
【Plot related】

Well, the third game of the Chess Saint Tournament is still a bit of a post-match final, so I’ll start writing about the story of the qualifying tournament tomorrow.

[Reality]

Regarding the relevant plot of this chapter, I feel it needs to be explained that it is indeed based on reality.

Whether the current number one Go player Shen Tiandi can beat the first generation of AlphaGo is still a controversial topic, but the mainstream view is that Shen has an advantage when he is not sick, or is good at winning. No one thinks that Shen Tiandi cannot beat the first generation of AlphaGo just because it is a deep learning AI - even if the two are very different in terms of computing power.

Simply put, in the game of Go, humans are now really catching up with the puppy's tail.



(End of this chapter)