top of page
Writer's pictureTakumigo

Cho Chikun's "My Resume" (29): Resistance to AI


original title:趙治勲 私の履歴書(29)AIに抵抗感

Original article: The Nikkei

Author: Cho Chikun, Honorary Meijin


In recent years, the most talked-about topic in the Go community has been the emergence of Go AI (Artificial Intelligence). In March 2016, the news that "AlphaGo" defeated South Korean top player Lee Sedol with a score of 4 wins and 1 loss had a huge impact on the Go world.


Personally, I believe that Lee Sedol’s defeat in that match was due to underestimating his opponent. When he saw the version from six months earlier, he thought, "I can beat AI like this," and thus, he underestimated it. If he had studied the latest version, perhaps he wouldn’t have lost.


In fact, in November of that same year, I played against a Go AI developed in Japan and won. I entered the match with the mindset of "I will definitely win," and the result was as I expected.


However, looking back, AI at that time still had some exploitable flaws. But in the following years, AI evolved remarkably, and it's said that now, even top professional players cannot win, even with a two-stone handicap.


Today, the top players in the world are South Korean Shin Jinseo and Chinese Ke Jie, both products of AI. However, I can no longer understand their games at all. Moves that were once considered bad shapes or absolutely unviable, they can use without hesitation. Their style of play is completely different from the Go thinking I’ve been studying all my life; it feels like I’m watching a Shogi match.


Go is a game about competing for territory, and for amateur players, it’s often difficult to tell who’s winning until the endgame calculation. Even for professional players, it’s not easy to judge who’s in a superior position midway through the game. Situation assessment is extremely difficult, and for a pessimistic player like me, I sometimes make moves when I’m in a favorable position, only to lose the game.


However, AI can draw conclusions in any situation. When we spend hours researching and still cannot judge the superiority of a position, AI can immediately calculate something like "Black’s winning rate is 70%." Humans have no countermeasure against this. It feels like another god—AI—has descended into this “game created by the gods,” and all that humans can understand is their own foolishness.


AI can instantly give conclusions whether it's morning or late at night, so it would be a shame not to use it for learning. I recommend that young people study through AI.


As for me, I do not plan to use AI. I don’t think that even if I started using AI now, I would be able to beat the younger top players. I can only compete using what I have learned so far. Even if someone were willing to install AI on my computer, I have no intention of using it.


In my 60-plus years of playing Go, I have never learned by looking at answers. Judging whether a move is good or bad, and whether the situation is favorable or not, and constantly struggling with these questions—that is the joy of Go. If that joy disappears, it feels like losing part of my soul. It's like someone saying to me, "Instead of spending time and effort drawing a picture, wouldn’t it be quicker just to take a photo?" I always feel some resistance to that.


On the other hand, through playing against younger players who have improved their skills using AI, I can indirectly absorb the benefits of AI, and that is enough for me.


At one time, my teacher, Mr. Kitani Minoru, taught me: "As a professional Go player, it’s shameful not to know how many points you won by when the game ends." I’ve always tried to follow this teaching, but at some point, I could no longer do it. Of course, this is partly due to age, and in my usual online practice games or research, I’ve already delegated the calculation tasks to the computer.


"So, shall I ask AI using my smartphone?" Someone as easygoing as me may, in the near future, really make full use of the smartphone I’ve never used before, and completely immerse myself in the world of AI.

25 views1 comment

1 comentário


Philippe Fanaro
Philippe Fanaro
17 de set.

AI isn't giving out absolute answers, just way stronger ones. It's like saying that an amateur shouldn't look at top pros games.

Curtir
bottom of page