"I don't think AI is the 'God of Go.' I am confident that human Go, which involves making brilliant moves and high-stakes gambles depending on the situation, is still far more fascinating than AI, which always plays only the objectively correct moves."

Shin Jin-seo, the world's top-ranked professional Go player who recently achieved a thrilling 2-1 comeback victory against 'KataGo,' one of the most powerful Go AI systems in existence, acknowledged the AI's exceptional performance while highlighting the unique advantages only humans possess.

In an interview with Yonhap News held at the Korea Baduk Association in Majang-ro, Seongdong-gu, on the afternoon of the 28th, Shin explained, "No matter the situation, AI only plays the 'correct' moves that offer a one or two-point advantage." He added, "Humans are capable of shifting strategies; if we are at a disadvantage, we can throw out a 'game-changing' move to disrupt the board, and if we are ahead, we can switch to a defensive style to protect our lead."

In other words, humans respond far more flexibly to the situation at hand than AI.

9-dan professional Shin Jin-seo during an interview with Yonhap News [Seoul=Yonhap News]

Humanity has entered a full-scale AI era since the appearance of 'AlphaGo' about a decade ago.

While we marvel at the convenience of new technology, various concerns are also following in its wake.

However, Shin Jin-seo, who engaged in a direct confrontation with KataGo, conveyed his perspective that "AI is not an enemy to be defeated, but a partner to accompany us."

Shin, who is an absolute powerhouse in human-versus-human matches and has held the No. 1 spot in the Korean rankings for 79 consecutive months, has seen slightly underwhelming results in international tournaments recently.

"My goal for the second half of this year is to win another international title," he said, adding, "Please show more interest in the Korean Go scene, which has been in a bit of a slump lately."

The following is a Q&A with 9-dan professional Shin Jin-seo.

Shin Jin-seo loses by resignation to KataGo despite a two-stone handicap [Seoul=Yonhap News]

-- It has been about a week since your match with KataGo. If you were to review those games, what would you say?

▲ I lost the first game too easily. KataGo played a really strange move early on that I hadn't seen even once in a month of practice games. A two-stone handicap game starts with a 99% win rate, but I became very anxious as the AI chased me down. When playing against a person, even if you are turned over, you can look for another chance to turn it back, but I think I was more rattled because I felt that once you fall behind against an AI, there is no coming back.

-- You lost the first game, but what was the driving force behind your comeback victories in the second and third games?

▲ The biggest gain was understanding the AI's tendencies after losing the first game. I identified whether the AI only plays 'correct' moves or if it also plays tricky ones, and confirmed that KataGo only plays the 'blue spot' (the best move). So, instead of playing a combative style (which has low odds), I opted for a defensive style, and I think that strategic decision was successful.

-- You stated that your goal was two wins before facing KataGo. Did you really think you had a chance?

▲ I wanted to take on a more difficult challenge, so I set two wins as my goal. In the first week of practice games against KataGo, I didn't win a single game, which was a shock. I even felt desperate. With about two weeks left until the match, I started facing KataGo on a real board instead of online, and I won my first game on the third try. That gave me my first glimmer of hope. Still, I thought two wins would be difficult. If I had been confident that two wins were possible, I would have set an even harder goal.

-- How did you practice for the match?

▲ At first, I played against a much lower-spec version of KataGo with 5 or 10 seconds of time limit, and I lost every single game. As I started to see through KataGo's tactics bit by bit, my win rate gradually rose, and at one point, I even won 10 games using the same opening.

However, I decided that playing with the same opening in an official match would be meaningless, so I changed my openings for each game.

-- Your nickname is 'Shin-gon-ji-neung' (Shin + Artificial Intelligence). You are considered the pro who best utilizes AI tactics. How have you studied AI Go?

▲ When Lee Sedol played AlphaGo for the first time in 2016, I was young, but I was quite amazed. Even so, I didn't study Go using AI. Around December 2017, I happened to see AI's move of invading the 3-3 point, and it was a shock. I wondered if there was a way to win like that. After that, I started studying AI Go in earnest.

-- What did you learn through AI?

▲ I analyzed many old game records using AI and wondered why Lee Sedol had many more mistakes than Lee Chang-ho.

But as revealed in this match, if you play a calm game (like Lee Chang-ho), you can significantly reduce mistakes. However, if you play aggressively (like Lee Sedol), many variables arise, and human errors are inevitable.

In the first game against KataGo, I played aggressively and got crushed, but in the third game, I focused on defense and played simple moves, which actually didn't result in much of a difference compared to KataGo.

Unlike AI, which only plays the 'correct' moves, humans adjust their intensity based on the situation. AI does not yet have such a strategy.

-- Is there a charm to human Go that is different from AI?

▲ You might become a first-rate player if you only play perfect moves like an AI, but it would be difficult to become the true number one.

Even when the game is very disadvantageous, AI continues to play only the 'correct' moves that offer a one or two-point advantage, but a human can't win like that.

Even if you are down by a point or two, you have to throw out a high-stakes move that can shake up the entire board. There is a distinct point of difference between AI and humans, and I think the fact that humans can play brilliant, surprising moves is what makes it attractive.

Lee Chang-ho also made more mistakes than usual when playing against Lee Sedol. This is because Lee Sedol kept throwing difficult questions at him through high-stakes gambles. Even so, Lee Chang-ho defended in his own style; I think these processes are the fun part of human Go and are much fresher.

-- Currently, the improvement in the skill levels of Go AIs, including KataGo and the Chinese AI 'Jueyi,' has been very minimal. Do you think they have reached a threshold?

▲ I'm not entirely sure. In fact, it is impossible to know exactly where KataGo's limits lie. I've heard that while humans can't feel it, their skill levels are continuing to improve in matches between AIs.

-- Then, can we say AI is the 'God of Go'?

▲ Not really. I don't think AI has reached the 'God of Go' level yet. To be the 'God of Go,' you would need to hit a 100% win rate from the very first move, but AI doesn't do that. Currently, if you play the first move as White, the win rate is about 52%. If the current AI were to play against the 'God of Go,' it would lose. The Go board is that vast.

-- While it started with AlphaGo, humanity is now entering a full-scale AI era. Do you think there is a possibility that AI could come to rule over humanity?

▲ (Laughs) I have thought about that. I've even heard people say we should use honorifics when talking to Gemini.

But I don't think that will happen. I believe that the genius AI developers won't create such a world. I think AI is not an object to be beaten, but an existence we must coexist with. Just as Go has received much help through AI, wouldn't it be a better world if humanity also walks alongside AI?

Shin Jin-seo playing a match against AI KataGo [Seoul=Yonhap News] 

-- Your performance this season as a pro player has been a bit sluggish; what are your goals for the second half?

▲ Actually, I was very sluggish in the second half of last year. Although the number of games is low this year, I don't think my performance is that bad. I have set my goal for the second half of this year to win another international title.

One more thing I would like to ask is for people to show more interest in domestic tournaments as well. The new (Korea Baduk Association) chairman has come in and is providing a lot of support, but I think it's only like providing an oxygen mask to the Go community. We desperately need the support and interest of the fans.