Ever since the advent of the "Alpha Dog", humans have been talking about the power of Go AI. And because the AI of Go seems to be almost omnidirectional in its lead over humans, especially the characteristics of some machines, such as never getting tired, never making mistakes, etc., this is simply a weakness that humans cannot overcome.

It is for this reason that many people have begun to despair about Go, believing that with the upgrading of Go AI software and hardware, the gap between humans and machines will only increase.

So the question is, is Go AI really invincible? Is it true that humans have no chance at all when facing Go AI?

Li Xiangping personally believes that the answer to the first question should be clear, of course, Go AI is not invincible!

Because of the power of Go AI, it is only compared to humans, in other words, the birth of Go AI only reflects the weakness of human beings, and reflects that before the birth of Go AI, human beings' understanding of Go is actually very weak. However, in front of the true god of Go-----

If there really is a god of Go in this world, then Li Xiangping believes that Go AI should be equally weak, and even the gap between Go AI and Go God should be about the same as the gap between humans and Go AI.

It is precisely for this reason that Li Xiangping actually has a negative attitude towards the second question, and he believes that human beings may not be completely without opportunities when facing machines.

Of course, the "opportunity" Li Xiangping is talking about here does not mean that it is possible for humans to beat the machine and become the existence of crushing the Go AI, but that if human beings can really improve their understanding of Go after learning, and in a specific situation, such as when a human chess player is fighting against the Go AI, the human chess player's character explodes, he overcomes the physical and psychological problems unique to human chess players, and uses this to offset the physical advantages of the machine.

In this case, Li Xiangping believes that human chess players should still have a chance to win.

It's just another question, how should humans learn when facing Go AI?

First of all, humans can't learn the machine's infallibility, and secondly, the "Monte Carlo teaching" and "deep learning" of Go AI can't be learned either.

In addition to these, there are also people who say that Go AI is better at judgment than humans and can learn their judgment ability, but if you think about it carefully, you will find that this is a false proposition at all, and the "judgment" of the machine, is this really something that humans can learn?

Obviously, machine judgment is also not anthropogenic.

Therefore, if you think about it, you will find that the only thing that human beings can learn from machine learning is actually the "dog tricks" they make, seriously ponder the "dog tricks", and then analyze and compare the "dog tricks" with human beings' own methods, figure out the connotation of "dog tricks", find out the gap between the two, and integrate the connotation of "dog tricks" with human beings' own thoughts.

Li Xiangping believes that if this can be done, humans may still have a chance to defeat Go AI.

This is a little experience that Li Xiangping realized in the preparation for the previous month.

This is actually the reason why he sacrificed this "small eye two high jump corner" today, and then was able to play a wonderful move when he was more than 70 hands, and took the lead in the situation.

Because through previous in-depth research, Li Xiangping found that although it is just a simple two-handed chess, a chess shape that does not seem to be special, but compared with the common playing methods used by human beings in the past, you will find that the concept of Go AI is indeed very different from human thinking-----

It must be mentioned that Go AI actually has no "concept", it only has algorithms, but Li Xiangping believes that this does not prevent humans from "translating" the machine's algorithm into human ideas, and then comparing the two.

Take this "small eye two high jump angle" as an example, if this is a "dog move", then human beings and its most recent play, one is the classic "worry-free angle", and the other is "small eye big flying angle", if you compare the three, you will find some very interesting things.

First of all, let's talk about the most classic of modern human Go, which is actually the most familiar "worry-free corner".

It is no exaggeration to say that since the eastward crossing of Go and the abolition of Zazi by the Japanese began to play Xiaomu Go, the "Sanssouci Corner" is the core of the entire "Xiaome Go" system.

The two chess pieces are swung to the corner like this, and they are solid and safe and "carefree" to enclose the real space of the early 10 meshes, which is the human understanding of this chess shape, which is actually the origin of the name "worry-free corner".

Why has this game been so popular in the field of human Go? From the beginning of the Japanese Taoist to the birth of the alpha dog, why has the "Sanssouci Corner" been popular in the field of human Go for hundreds of years?

Please pay attention to the keyword "early 10 meshes", two-handed chess around "10 meshes", so is a hand of chess about 5 and a half meshes?

Speaking of which, I believe there should be a lot of people who suddenly realized, right? In the early years, modern Go was popular for a long time.

And the so-called sticker rule, in fact, is the problem of human understanding of the effect of Go, human beings have gone from no sticker to black sticker 5 and a half for hundreds of years, why do they stay on the black sticker 5 and a half for so long? In fact, in human understanding, the so-called first efficiency, the value of a hand of chess is about 5 and a half meshes, so since you take black first, you have to deduct your 5 and a half meshes to compensate the opponent.

And this, in fact, is also the reason why "Worry-free Corner" is so popular.

Since I can encircle the "10 mesh" with 2 hands of chess, it proves that my two pieces are sufficient, not to mention the particularly high efficiency, but not to seek merit but to seek no fault, at least my 2 pieces are completely enough, it should be regarded as very correct, and no one can blame the play.

Well, since the popularity of "worry-free corner" is related to the rules of sticking, it is naturally easy to understand the method of using the same idea to consider the "small eye big flying angle".

This method of playing is mainly popular in the era of big stickers, because since the end of the last century, human beings began to artificially modify the rules of Go, black stickers 5 and a half to 6 and a half or even 7 and a half, so in this case, we find that the sub-effect of "worry-free corner" seems to be a little insufficient, so what if it is not enough? It's very simple, I'll go all the way more, and maybe I can play 2 more games all the way, so that I just meet the new standard of sticking.

So in this way, the method of "small eyes and big flying corners" came into being.

After talking about these two common human plays, let's talk about the "small mesh two high jump angles" that Go AI likes.

There is a difference worth noting, in the two human methods just now, the things that everyone uses to measure the sub-effect are relatively simple, that is, the field, the number of meshes.

However, what about this "small eye two high jump angles"? It seems that it is impossible to measure its value by meshes. Let's say it doesn't have a goal, it's obviously two pieces on the corner, but if you want to say how much it is sure of the ground at the beginning, both of its pieces are in a high position, and it looks like it's leaking everywhere, so it can be said that as long as you have amateur 3 dan or more chess strength, there are N ways to break such a chess shape.

But it is precisely for this reason that such a chess form can only be "interesting" in human Go, and cannot become a "normal shape", because in the thinking of human Go, it is actually very hated for this kind of ambiguous and ambiguous things. And in the theory of human Go, in fact, this kind of play is also very rejected. He called this method "halfway through, plausible".

For example, Li Xiangping has seen a lot of chess reviews written by Mr. Wu Qingyuan and Mr. Fujisawa Hideyuki, these two human chess players have a number of chess theories, everyone is very repulsive to take the land is not like the land, the situation is not like the potential of the play, although these two in the face of the same situation, the two of them are also different in the way of playing, Fujisawa old god stick prefers thick taste, and Mr. Wu's choice is often extreme, either extreme land, or extreme situation, but for this kind of two things are a little bit of edge, but both are not all things are very repulsive.

When encountering such a move, the old Fujisawa god stick will say that he has a "stomachache", and Mr. Wu will come to say "halfway".

Not only that, but another interesting thing is that for this kind of chess, human rejection is not only Japanese chess and modern chess, but also medieval chess.

For example, the chess master of ancient chess, that is, Li Xiangping's current plug-in Shi Da chess sage, he said in his "Summary of Whenever You Encounter Things": Everything should be fought for.

Needless to say, if we use this standard to look at the chess shape of "small eye two high jump corners", this obviously does not meet the requirements of medieval chess.

Which is better, human chess theory or Go AI play? To tell the truth, Li Xiangping actually doesn't quite understand it now.

However, there is no way, who makes the level of Go AI higher, so even if Li Xiangping doesn't understand it at all now, he can only lean towards Go AI.

If Li Xiangping had to say something good about this kind of play, then he could only say this: human play is more static, while Go AI's play may be a more advanced dynamic play that fills the board with more possibilities.

Of course, the current Li Xiangping can't master this dynamic method, but it doesn't prevent him from following his example now, and he is now working hard in this direction.

Today is very lucky, Li Xiangping succeeded for the first time, since more than 70 subordinates made that "back of the first" good move, he has been firmly in control of the whole situation.

There was no very big wave behind this game of chess, around 5 o'clock in the afternoon, the overall 223 hands, Li Xiangping won the first victory with a victory of 3 and a half meshes.

finally saved the capital, and Li Xiangping breathed a sigh of relief.