Shin Jinseo 9p is the number one player in South Korea and many consider him the best human go player in the world now. He quickly rose to the top after AI became a major part of the go world. Shin is such a devoted student of AI that his nickname is “AI Shin”. Everyone at the top level of the pro world now has to learn as much of the AI opening moves as possible, that’s through about the first fifty moves. And not only the moves, but the winning percentages associated with them. Shin Jinseo’s first major win on the international level was the LG Cup in 2020 where he defeated Park Junghwan 9p in the finals. At that time, Park Junghwan was considered one of the very top players. Shin also won the Chunlan Cup in 2021, defeating Tang Weixing 9p of China. He played in a special seven game match against his fellow countryman Park Junghwan in 2021 and managed to win all seven games against the player who had been considered the best in Korea. His winning percentage has been very high as a pro, over 70%, recently he has reached 90%. These three games reflect his playing style, which is heavily dependent on AI in the opening and tends to be relatively aggressive.
This game was played in the Chinese City League A, in which Shin Jinseo 9p is one of the invited foreign players, on June 8, 2021. Yang Dingxin is currently ranked number two in China, behind Ke Jie 9p. Shin Jinseo is Black and Yang Dingxin is White. Under Chinese rules the komi is seven and a half points. The basic time is two hours each. As is common nowadays, we will see a lot of AI influence in the opening.
White’s nirensei is the most popular pattern among pros and AI. Normally, Black would now defend the lower right corner with an extension.
However, in this case, Black plays the early 3–3 invasion that has been popularized by AI. The Chinese pros have exclusive access to the best AI program, which is called Fine Arts (and this is the program I have in mind whenever I mention AI). This move has the highest winning percentage in AI analysis.
Blocking on this side makes sense for White, since he has the most potential on that side. Pros often opt to block at 7 with 6, focusing on reducing the opponent’s potential, but I do not recommend that approach for weaker players.
At A, seeking sente, is the most popular move for 8. The game play is rather complicated.
Black chooses the most complicated variation with 9. Shin Jinseo lives up to his nickname of “AI Shin” here. He’s known for his deep study of AI variations.
If White now hanes at 1, this is the most complicated variation. After 2, White pulls back at 3. After 4 and 5, Black cuts at 6. There are many complicated variations after this. I would not encourage weaker players to try this variation. Even strong players make mistakes dealing with this situation.
Yang Dingxin decides not to challenge AI Shin in this situation and chooses to play solidly with 10. Of course, Black then extends to 11. Normally, since Black gets out, this is seen as a good result for Black, but White does get sente.
White approaches the 3–4 stone corner with 12, and Black makes the response now seen as best by AI at 13. This makes Black’s corner very strong, but it puts no pressure on White.
So White takes away Black’s corner with 14. We have learned from AI that Black can block on either side in this case, though the traditional view is that Black should block at A since he has more potential in that direction.
If White now attaches at A, Shin Jinseo is of course ready to initiate the complicated AI version of this joseki by playing at B.
But Yang Dingxin again avoids the more complicated version. He continues to play solidly with White 20. Yang does not play as aggressively as Shin Jinseo.
In the past, letting Black get this outside thickness was seen as good for Black, but AI doesn’t support that idea. The key issue now is how Black can use this thickness. White has sente.
Yang plays 24 to prevent a black play there that would give Black a nice moyo.
We might have expected “AI Shin” to go into the corner with 25. A normal response for White now would be at A.
Normally, we would expect a pincer at A with 26.
It would also seem normal for Black to extend to B after 26, which is certainly playable, although that would give White sente.
However, Black attacks at 27 instead, aiming for a moyo. This is another AI approved move.
Again, we might expect White to play 28 at A, pincering the black stone.
Black extends, looking at the moyo possibility. Normally, White would respond by protecting the cutting point with A, but if White does that, Black would turn at B, developing tremendous right side and center potential.
So White extends at 30, hoping Black will extend at A to prevent a hane at the head of three stones, which would allow White to then play at B.
However, Black doesn’t follow White’s plan and aggressively cuts with 31. This starts a complicated fight. Separating and connecting always play a major role.