I recently read 먼저 온 미래 (roughly, "The Future That Arrived Early") by Chang Kang-myoung. The book traces how the world of Go changed in the decade after AlphaGo's 2016 victory over Lee Sedol, one of the world's top Go players.
AlphaGo and the team behind it at Google DeepMind have been covered plenty: documentaries, articles, interviews. But not much has been said about what happened to the Go world after. That's what makes this book so interesting. More than the match itself, what stayed with me was everything that followed.
Here are a few changes from the book that stood out.
AI became the final authority
Go has too many possible variations for humans to fully grasp. That's why professional players have long relied on more abstract language to explain what was happening on the board: influence, thickness, momentum, board sense, awkward shape, a bad aftertaste, and so on. Some players went even further. They talked not just about winning and losing, but about balance, elegance, and beauty. For them, Go was closer to art.
AlphaGo upended all of that. As AI began offering clearer answers through win rates and optimal sequences, the old language of intuition quickly lost its weight. Long-established game records and joseki, the standard opening sequences of Go, no longer carried the same authority. Once AI could reveal stronger play than any human game record or theory, it stopped being one source of insight among many and became the standard for judging moves.
Before AlphaGo, people proposed new ideas, argued over them, and let different styles and interpretations coexist. After AlphaGo, that order changed almost overnight. Game records, joseki, commentary, and intuition were no longer the final word. AI was.
For professional players, it was more than a loss
For professional players, AlphaGo's victory didn't feel like just another loss. Go was more than a job. It was the world they'd built their lives around, shaped by instinct, discipline, and pride. That's why so many of their reactions sounded closer to grief than disappointment. "It felt like everything was falling apart." "It felt like the world I knew had collapsed." "What value was there in all the effort I had put in?" Those aren't the words of someone who simply lost to a stronger opponent.
Lee Sedol eventually retired and put it this way: "When I was young, I was taught that Go was a kind of art. I think of it as a work of art made with stones, but what kind of art can it be now? The art I learned has collapsed. I felt it would be hard to keep going." What broke after AlphaGo wasn't just a few patterns or ideas. It was the meaning these players had spent their whole lives building around the game.
At first, some people tried to frame the loss differently. Since AlphaGo had been trained on human game records and centuries of human knowledge, you could still say this wasn't a machine simply surpassing humans, but human knowledge continuing in a different form.
Then AlphaGo Zero arrived in 2017 and even that interpretation stopped working. AlphaGo Zero wasn't trained on human games. It was given only the rules of Go and improved by playing against itself over and over. In just 36 hours, it surpassed the version of AlphaGo that had beaten Lee Sedol. After 72 hours, it went 100–0 against that earlier version. For many professional players, that may have landed even harder than Lee Sedol's defeat itself. It suggested that everything humans had built up over thousands of years wasn't, after all, essential to playing the best possible Go. It may even have looked like a set of human limitations that had to be left behind.
Players started training to play more like AI
Before AlphaGo, professional study was mostly about reviewing the games of stronger players, discussing them with other professionals, and making sense of them in your own way. Even when two players studied the same game record, what they took from it could be completely different. After AlphaGo, that changed fast. Study became about checking where your judgement diverged from AI's recommendations and trying to close that gap. Deepening your own interpretation mattered less. Playing closer to AI mattered more.
Professional players began reviewing games with AI analysis tools. These programs calculate win probability move by move and mark the best point with a blue dot, which players call the "blue spot". The problem is that AI doesn't really tell you why that move is best. So studying turned into trying to understand the move AI chose, then adjusting your own play in that direction. Players who resisted AI-style play quickly fell behind. If you wanted to stay competitive, you had to train yourself to play more like AI.
The clearest example of this shift is Shin Jinseo. He matters because he became the world number one in the AI era, and because of how he did it. His move choices lined up with AI recommendations more closely than most top players. He pushed harder than anyone to close the gap between AI's judgement and his own. Shin himself said this kind of study wasn't enjoyable and was mentally draining, but that he'd worked harder than anyone. After AlphaGo, the standard for being the best changed. It was no longer about pushing your own style to the limit. It was about endlessly closing the gap between your own judgement and AI's. Shin Jinseo became the number one player in that new world.
The authority of players, teachers, and commentators all weakened
What changed after AlphaGo wasn't just the confidence of individual players. The authority professionals had long held started to erode too. Professionals used to be seen as people who could read deeper and judge more accurately than everyone else. Once AI arrived, that information gap narrowed very quickly. The shift showed up early in the teaching market. Go schools, private lessons, and teaching games lost value, and the assumption that professionals were closer to the right answer started to weaken.
Teachers were affected too. In the past, if a teacher said, "This is how you should play here," that was often the end of it. Now even students can check with AI straight away. Teaching shifted from declaring the right answer to comparing your own thinking with AI's and explaining the gap. The teacher became more of an interpreter than an authority.
Commentators changed as well. AI win-rate graphs made judging the flow of a game easier, but they also weakened the commentator's role. The commentator was no longer the person best placed to say who was ahead. More often, they were explaining the judgement that was already visible on the screen. Watching Go also changed. It shifted towards live evaluation. A professional move was no longer just something to admire or discuss. It was also something anyone could instantly see as a mistake.
The same thing happened to players, teachers, and commentators alike. They were no longer the people who had the answers. Expertise stopped meaning you had privileged access to the truth. It meant you could explain the conclusions AI had already reached.
In some ways, the world of Go became more open
After AlphaGo, the world of Go became colder and more standardised. But in another sense, it also became more open. The Go world had always been more closed than it looked from the outside. As Lee Sedol once recalled, there was even a time when you had to physically go to the Korea Baduk Association and photocopy game records. People with quicker access to high-level game records, better teachers, or stronger training environments had a real advantage. Players outside the major Go countries were often at a disadvantage in that system.
AI broke some of that open. In the past, even when you learned from a stronger player, the explanation could be vague, and it was often hard to tell whether you were really on the right track. AI, at least, could show you more directly which move was better and which direction looked more promising. Studying may have become more sterile, but the path also became shorter and clearer. You no longer had to rely as heavily on instinct, connections, or environment to improve quickly.
One of the most interesting examples was the rise of women players. Go had long been male-dominated, and that bias shaped everything: training schools, institutions, mentorship, personal networks. With that in mind, the narrowing gap between male and female players after AI became part of training feels especially significant. It suggests that the old gaps in Go were not just about talent. They were also shaped by access to information and training, and by the informal structures around the game. AI made Go harsher in some ways, but it also opened up parts of the game that had once been more closed off.
So the post-AlphaGo world became more uniform and less romantic, but also more open in important ways. AI took some of the mystique out of Go, but it also widened access to the game itself.
What happened to Go ten years ago was, in many ways, a special case. It was as if AGI had arrived almost without warning, and professionals had no choice but to accept it. That makes Go hard to compare with what is happening in other industries now. As a sport and a game, Go could carry on even after AI surpassed all human players, much as chess did after Deep Blue.
Even so, the way Go changed is worth thinking about. What happens when something beyond human ability arrives almost overnight? How does old authority lose its grip, and what starts to matter in its place? This book raises questions that feel especially relevant right now.