人工智能正重塑顶尖围棋棋手的思维方式。
内容总结：
人工智能重塑造棋思维：十年人机大战后，围棋世界如何被AI重塑？
十年前，谷歌DeepMind研发的人工智能程序AlphaGo击败韩国棋手李世石，震惊世界。如今，人工智能已彻底改变了这项拥有2500多年历史的棋类运动，不仅颠覆了传承数百年的棋理，更重塑了职业棋手的训练方式与思维方式。
在首尔韩国棋院，昔日棋手对弈时棋子落入木盘的清脆声响，已被鼠标点击和屏幕蓝光所取代。棋手们紧盯着AI程序（如目前主流的KataGo）给出的胜率分析与推荐选点，反复拆解棋局。世界排名第一的申真谞每天清晨都会打开KataGo进行训练，他的棋风因高度模仿AI而被戏称为“申智能”。研究表明，他的行棋与AI推荐的吻合率高达37.5%，远超职业棋手28.5%的平均水平。
棋道革新：从“艺术创造”到“高效学习”
AI带来了根本性的棋理革命。许多曾被视作常识的定式与布局原则被推翻，全新的招法不断涌现。职业棋手的训练核心，从创造性的自我探索，转向尽可能精准地复现AI的选点。布局阶段尤其明显，前50手棋日益“同质化”，棋手个性表达的空间被压缩。李世石感叹，当棋步从“答案库”中复制时，围棋已不再是他曾视为“艺术”的追求。
人机共生：新挑战与新机遇
尽管AI的决策如同“黑箱”，其思考维度对人类而言仍显神秘，但它也成为了一个“民主化”的超级教练。尤其对长期处于竞争劣势的女棋手而言，AI提供了前所未有的高质量训练环境，打破了以往以男性棋手为核心的传统训练壁垒。顶尖女棋手金彩瑛等人借助AI分析，打破了对于顶尖男棋手“不可战胜”的心理敬畏，近年来女子棋手战绩显著提升。
未来之路：人类棋手的价值何在？
尽管AI在技术层面已远超人类，但观众依然更爱观看人类对弈——充满个性、失误与逆转的棋局才更具情感张力。申真谞认为，AI是导师、伙伴和北极星，让他保持谦逊并不断精进。而退役后的李世石则对围棋的未来抱有一丝新的期待：或许在AI的辅助下，人类棋手终能下出一盘无限接近完美的“传世名局”。
如今，职业围棋已离不开AI。它既被指责“扼杀创意”，也被视为提升整体水平的工具。在人工智能的浪潮中，围棋这项古老艺术正在寻找属于人机共生时代的新身份与新故事。
中文翻译：
人工智能正在重塑顶尖围棋棋手的思维方式
十年前那场里程碑式的胜利之后，人工智能如今已主导围棋训练。棋手们正在思考这对围棋这项运动意味着什么。
在首尔东部弘益洞幽静的住宅区巷弄深处，矗立着一座石瓦斑驳的建筑，门牌上印着"韩国棋院"——韩国职业围棋的最高管理机构。围棋这项古老技艺在韩国享有神圣地位。
但如今，建筑内部曾经回荡着棋手探手入棋钵的轻柔声响的房间，现在充斥着鼠标点击声。棋手们俯身于显示器前，在人工智能程序中复盘对局；另一些人则围聚在棋盘旁，争论下一步的最佳走法，而教练则汇报他们的选择与AI建议的契合度。还有人静坐一旁，观看AI程序之间的对弈。
十年前，谷歌DeepMind的AI程序AlphaGo击败韩国棋手李世石，震惊世界。此后数年，人工智能彻底颠覆了围棋。它推翻了延续数百年的最佳行棋原则，引入了全新的策略。如今棋手训练的目标是尽可能复现AI的招法，而非创造自己的着数——即便他们仍无法完全理解机器的思考逻辑。在当今棋坛，不使用AI几乎不可能参与职业竞争。有人认为这项技术扼杀了围棋的创造性，也有人认为人类仍有创新空间。与此同时，AI正在使训练资源民主化，更多女性棋手因此崭露头角。
对世界排名第一的棋手申真谞而言，AI是无价的训练伙伴。每天清晨，他都会打开名为KataGo的程序。因行棋风格与AI高度相似而被戏称为"申智能"的他，会追踪屏幕上代表AI推荐最佳落点的"蓝点"，在数字棋盘上调整棋子位置，试图理解机器的思考逻辑。"我不断思考AI选择某步棋的原因，"他说。
备战比赛时，申真谞每天清醒时间大多沉浸于KataGo研究。"这近乎苦修，"他坦言。据韩国围棋联赛2022年研究，申真谞的行棋与AI建议的吻合度达37.5%，远高于研究统计的棋手平均28.5%的吻合度。
"我的棋风改变很大，"申真谞表示，"因为必须在某种程度上遵循AI指引。"韩国棋院透露，为纪念AlphaGo战胜李世石十周年，已联系谷歌DeepMind筹划申真谞与AlphaGo的对决。谷歌DeepMind发言人表示暂无法提供相关信息。但若比赛成真，使用更先进AI程序训练的申真谞对获胜持乐观态度："当时的AlphaGo仍有缺陷，针对弱点进攻我认为可以取胜。"
人工智能重写围棋法则
围棋是2500多年前起源于中国的抽象策略棋盘游戏。双方在19路棋盘上交替落子，通过围困对方棋子争夺领地。这项运动具有惊人的数学复杂度——约10¹⁷⁰种可能棋局配置远超宇宙原子总数。如果说国际象棋是战役，围棋就是战争：你既要在角落扼杀对手，又需在另处抵御入侵。
训练AI下围棋需向神经网络输入海量人类棋谱，这种计算系统模仿人脑神经元网络。战胜李世石后更名为AlphaGo Lee的程序，通过3000万步人类棋谱训练，并经过数百万自我对弈完善。2017年，其继任者AlphaGo Zero从零开始学习围棋。它不研究任何人类棋谱，仅通过自我对弈基于规则学习。这种白板式方法因不受人类认知局限而更强大，训练三天后即以100:0击败AlphaGo Lee。
谷歌DeepMind同年让AlphaGo退役。但随后涌现出受AlphaGo Zero启发的开源模型浪潮。如今KataGo已成为韩国职业棋手最广泛使用的程序，它比AlphaGo更迅捷精准，不仅能预测胜负，还能实时判断棋盘每点归属。AlphaGo Zero通过局部观察拼凑对棋盘的理解，而KataGo学会通览全局，对长期策略形成更佳判断——它不仅学习取胜，更追求最大化得分。
软件重塑行棋方式
数百年来，职业棋手通过发展启发式方法替代蛮力计算，驾驭围棋的天文级复杂度。优雅的开局策略在空棋盘上构建抽象秩序，早期侵入角落曾被视作亏本买卖。历代棋手不断为棋理宝库增添新原则。
但"AI改变了一切，"韩国围棋解说员朴正祥指出，"曾经的基本常识如今完全不被采用，前所未有的新技术却流行起来。"
最显著变化体现在开局。围棋始于空白棋盘，前50手本是抽象思维与创造力的画布，棋手在此镌刻个性与哲学：李世石善用挑衅招法制造混乱；2017年败于AlphaGo Master的中国棋手柯洁以灵动想象令人目眩。如今棋手们背诵着AI推荐的同一套高效计算型开局，博弈重心转移至中盘——这里原始计算比创造力更重要。
AI训练导致棋风同质化。柯洁曾哀叹观看重复开局的疲惫："我和观赛棋迷感受完全相同，观看过程非常累人痛苦。"棋迷为棋手打破套路而欢呼，但这样的时刻日益罕见。2023年研究显示，顶尖棋手超三分之一落子复现AI推荐。多位棋手坦言，如今每局前50手常与AI建议如出一辙。
"围棋已成为智力运动，"2016年败给AlphaGo三年后退役的李世石说，"AI出现前，我们追求更高境界。我曾将围棋视为艺术，但若从参考答案抄袭招法，那便不再是艺术。"
有棋手认为，下棋不再是开拓新疆域，而是遵循超人类神谕的指示。"我曾通过推进围棋技术、呈现新范式来激励棋迷，"李世石说，"但我下棋的理由已经消失。"
神秘的思维
留在棋坛的棋手正尝试重塑技艺，但厘清新原则并非易事。
世界顶尖女棋手金彩瑛身形纤巧却沉着非凡，自幼随职业棋手父亲学棋。当AI开始重塑围棋时，她发现自己需要重新开始。"我需要时间抛弃以往所学，"金彩瑛边说边将光标指向KataGo的蓝点与我共享屏幕，"多年积累的直觉原来都是错的。"
当她贴近显示器，闪烁的屏幕显示每步棋的胜率概率，却无任何解释。即使如金彩瑛、申真谞这样的顶尖棋手也无法理解AI所有行棋逻辑。"它仿佛在更高维度思考，"她说。向AI学习时，"重点不是理性推演每步棋，而是培养直觉——种本能感受。"
研究人员正试图破译围棋AI程序中编码的超人类知识以供人类学习。2024年，谷歌DeepMind研究人员从AlphaGo Zero的通用版本AlphaZero中提取新国际象棋概念，通过棋谜传授给特级大师。芝加哥丰田技术研究所计算机科学家尼古拉斯·汤姆林指出，目前棋手从AI系统领悟的围棋概念"可能只是潜在可学知识的冰山一角"，他曾参与研究AlphaGo Zero编码的围棋概念。
但提取这些知识仍困难重重。"顶尖棋手尚未能推演出AI行棋背后的通用原则，"明知大学围棋教授南治亨表示。尽管棋手能模仿AI招法，但由于其推理如同黑箱，他们仍未提炼出新范式。围棋可能正处在认知的过渡期。
民主化的导师
即便AI是位不透明的老师，却是位民主的导师。它为长期处于弱势的女性棋手注入了训练动能。南教授指出，数十年来，训练意味着跟随顶尖男棋手学习，最激烈的竞争发生在女性难以突破的男性圈层。"女棋手从未获得那种经验，但现在她们可以通过AI研习，训练环境变得有利得多。"更广泛而言，AI通过帮助所有棋手完善开局，缩小了棋手间的差距。
近年来女性棋手排名因此攀升。2022年，当时世界排名第一的女棋手崔精成为首位闯入国际围棋大赛决赛的女性。以"女斗士"著称的她与申真谞对决虽败犹荣，为女性围棋开辟新天地。2024年，金彩瑛赢得韩国围棋联赛季后赛成为头条新闻，她是该赛事唯一女棋手。
AI训练赋予金彩瑛新的自信。通过AI分析男棋手对局，打破了他们不可战胜的光环。"过去我无法衡量顶尖男棋手有多强——他们仿佛无敌。现在我知道他们也会犯错，招法并非总是精妙，"她说，"AI打破了心理屏障。"
棋手寻找新定位
尽管AI围棋水平远超人类，棋迷仍更爱观看人类对弈。"观看AI程序对弈对棋迷来说不太有趣，"解说员朴正祥表示，这类对局过于复杂难懂，过于完美反而缺乏激情。
棋手可模仿AI开局，但在棋盘可能性呈指数级增长的中盘，仍需依靠自身判断。棋迷沉醉于观看棋手失误与逆转，每颗棋子都彰显个性：申真谞棋风凌厉却带着机器般的沉稳；金彩瑛能巧妙驾驭最混乱的棋局。
"围棋中每步都是选择，对手以选择回应，"27岁的围棋爱好者、业余棋手金大辉说，"观看这个过程本身就充满乐趣。"
正是因为有金大辉这样的棋迷关注，申真谞在棋局中找到了意义："我能下出讲述人类独有故事的围棋。"
退役后，李世石寻找着能发挥人类优势的新领域。他开始设计棋盘游戏、发表演讲、在大学授课。"我正在寻找能享受并擅长的新领域，"他说。
但最近，他对离开的围棋运动重燃希望。"每位棋手的梦想都是下出名局，"他说——那种技术精湛、毫无失误、势均力敌的巅峰对决。"这就像海市蜃楼，"李世石轻笑，"也许AI能帮助我们下出名局。"
申真谞希望自己能实现这个目标。对他而言，AI是导师、伙伴，也是北极星。"我可能是最强的人类棋手之一，但有了AI，我不敢傲慢，"他说，"AI给了我不断进步的理由。"
英文来源：
AI is rewiring how the world’s best Go players think
Ten years after a milestone victory, AI now dominates Go training. Players are figuring out what that means for the game.
Burrowed in the alleys of Hongik-dong, a hushed residential neighborhood in eastern Seoul, is a faded stone-tiled building stamped “Korea Baduk Association,” the governing body for professional Go. The game is an ancient one, with sacred stature in South Korea.
But inside the building, rooms once filled with the soft clatter of hands dipping into wooden bowls of stones now echo with mouse clicks. Players hunch over their monitors and replay their matches in an AI program. Others huddle around a Go board and debate the best next move, while coaches report how their choices stack up against the AI’s. Some sit in silence, watching AI programs play against each other.
Ten years ago AlphaGo, Google DeepMind’s AI program, stunned the world by defeating the South Korean Go player Lee Sedol. And in the years since, AI has upended the game. It’s overturned centuries-old principles about the best moves and introduced entirely new ones. Players now train to replicate AI’s moves as closely as they can rather than inventing their own, even when the machine’s thinking remains mysterious to them. Today, it is essentially impossible to compete professionally without using AI. Some say the technology has drained the game of its creativity, while others think there is still room for human invention. Meanwhile, AI is democratizing access to training, and more female players are climbing the ranks as a result.
For Shin Jin-seo, the top-ranked Go player in the world, AI is an invaluable training partner. Every morning, he sits at his computer and opens a program called KataGo. Nicknamed “Shintelligence” for how closely his moves mimic AI’s, he traces the glowing “blue spot” that represents the program’s suggestion for the best next move, rearranging the stones on the digital grid to try to understand the machine’s thinking. “I constantly think about why AI chose a move,” he says.
When training for a match, Shin spends most of his waking hours poring over KataGo. “It’s almost like an ascetic practice,” he says. According to a study in 2022 by the Korean Baduk League, Shin’s moves match AI’s 37.5% of the time, well above the 28.5% average the study found among all players.
“My game has changed a lot,” says Shin, “because I have to follow the directions suggested by AI to some extent.” The Korea Baduk Association says it has reached out to Google DeepMind in the hopes of arranging a match between Shin and AlphaGo, to commemorate the 10th anniversary of its victory over Lee. A spokesperson for Google DeepMind said the company could not provide information at this time. But if a new match does happen, Shin, who has trained on more advanced AI programs, is optimistic that he’d win. “AlphaGo still had some flaws then, so I think I could beat it if I target those weaknesses,” he says.
AI rewrites the Go playbook
Go is an abstract strategy board game invented in China more than 2,500 years ago. Two players take turns placing black and white stones on a 19x19 grid, aiming to conquer territory by surrounding their opponent’s stones. It’s a game of striking mathematical complexity. The number of possible board configurations—roughly 10170—dwarfs the number of atoms in the universe. If chess is a battle, Go is a war. You suffocate your enemy in one corner while fending off an invasion in another.
To train AI to play Go, a vast trove of human Go moves are fed into a neural network, a computing system that mimics the web of neurons in the human brain. AlphaGo, which was later christened AlphaGo Lee after its victory over Lee Sedol, was trained on 30 million Go moves and refined by playing millions of games against itself. In 2017, its successor, AlphaGo Zero, picked up Go from scratch. Without studying any human games, it learned by playing against itself, with moves based only on the rules of the game. The blank-slate approach proved more powerful, unconstrained by the limits of human knowledge. After three days of training, it beat AlphaGo Lee 100 games to zero.
Google DeepMind retired AlphaGo that same year. But then a wave of open-source models inspired by AlphaGo Zero emerged. Today, KataGo is the program most widely used by professional Go players in South Korea. It’s faster and sharper than AlphaGo. It’s learned to predict not just who will win, but also who owns each point on the board at any given moment. While AlphaGo Zero pieced together its understanding of the board by looking at small sections, KataGo learned to read the whole board, developing better judgment for long-term strategies. Instead of just learning how to win, it learned to maximize its score.
The software has reshaped how people play. For hundreds of years, professional Go players have navigated the game’s astronomical complexity by developing heuristics that replaced brute calculation. Elegant opening strategies imposed abstract order on the empty grid. Invading corners early was a bad bargain. Each generation of Go players added new principles to the canon.
But “AI has changed everything,” says Park Jeong-sang, a South Korean Go commentator. “Fundamental moves that were once considered common sense aren’t played at all today, and techniques that didn’t exist before have become popular.”
The starkest shift has been in opening moves. Go starts on a blank grid, and the first 50 moves were canvases for abstract thinking and creativity, where players etched their personalities and philosophies. Lee Sedol fashioned provocative moves that invited chaos. Ke Jie, a Chinese player who was defeated by AlphaGo Master in 2017, dazzled with agile, imaginative moves. Now, players memorize the same strain of efficient, calculated opening moves suggested by AI. The crux of the game has shifted to the middle moves, where raw calculation matters more than creativity.
Training with AI has led to a homogenization of playing styles. Ke Jie has lamented the strain of watching the same opening moves recycled endlessly. “I feel the exact same way as the fans watching. It’s very tiring and painful to watch,” he told a Chinese news outlet in 2021. Fans revel when a player breaks from the script with offbeat moves, but those moments have become rarer. Over a third of moves by the top Go players replicate AI’s recommendations, according to a study in 2023. The first 50 moves of each game are often identical to what AI suggests, many players say.
“Go has become a mind sport,” says Lee Sedol, who retired three years after his 2016 defeat to AlphaGo. “Before AI, we sought something greater. I learned Go as an art,” he says. “But if you copy your moves from an answer key, that’s no longer art.”
Playing Go is no longer about charting new frontiers, some players say, but about following the dictates of a superhuman oracle. “I used to inspire fans by advancing the techniques of Go and presenting a new paradigm,” says Lee. “My reason for playing Go has vanished.”
A mysterious mind
The players who have stayed in the game are trying to reinvent their craft. But it can be hard to discern what the new principles are.
Disarmingly slight and formidably calm, Kim Chae-young, one of the top female Go players in the world, grew up learning the game from her father, who was also a professional Go player. But when AI began to reshape the game, she found herself starting over. “I needed time to abandon everything I had learned before,” says Kim who shared her screen with me as she pointed her cursor to the blue spots suggested by KataGo. “The intuition I had built up over the years turned out to be wrong.”
As she leaned close to her monitor, her blinking screen showed the winning probabilities of each move, with no explanations. Even top players like Kim and Shin don’t understand all of AI’s moves. “It seems like it’s thinking in a higher dimension,” she says. When she tries to learn from AI, she adds, “it’s less about rationally thinking through each move, but more about developing a gut feeling—an intuition.”
Researchers are trying to discover the superhuman knowledge encoded in game-playing AI programs so that humans can learn it too. In 2024, researchers at Google DeepMind extracted new chess concepts from AlphaZero, a generalized version of AlphaGo Zero that can also play chess, and taught them to chess grandmasters using chess puzzles. The Go concepts that players have picked up from AI systems so far are “probably only a small portion of what you could potentially learn,” says Nicholas Tomlin, a computer scientist at Toyota Technological Institute at Chicago, who coauthored a study probing Go concepts encoded in AlphaGo Zero.
But extracting those lessons remains a struggle. “Top-tier players haven’t yet been able to deduce the general principles behind AI moves,” says Nam Chi-hyung, a Go professor at Myongji University. Although they can emulate AI’s moves, they have yet to glean a new paradigm for the game because its reasoning is a black box, she says. Go may be in an epistemic limbo.
Even if AI is an opaque teacher, it’s a democratic one. It has supercharged training for female Go players, who have long been underdogs of the game. For decades, training meant studying under top male players, and the most competitive matches took place in male circles that were difficult for women to break into, says Nam. “Female players never had access to that experience,” she says. “But now they can study with AI, which has made their training environment much more favorable.” More broadly, AI has narrowed the gap between players by helping everyone perfect their opening moves.
Female players have climbed the ranks over the last few years as a result. In 2022, Choi Jeong, then the top female player in the world, became the first woman to reach the finals of a major international Go tournament. Dubbed “Girl Wrestler” for her fierce, combative style of play, she took on Shin. She lost, but the match broke new ground for women in Go. In 2024, Kim made headlines for winning the Korean Go League’s postseason playoffs. She was the only female player in the tournament.
Training with AI has given Kim newfound confidence. Analyzing male players’ moves with AI has shattered their veneer of infallibility. “Before, I couldn't gauge just how strong top male players were—they felt invincible. Now, I know that they make mistakes, and their moves aren’t always brilliant,” she says. “AI broke the psychological barrier.”
Go players find a new identity
Although AI has mastered Go far better than any player, fans continue to prefer watching people play. “A Go game between AI programs is not very fun for fans to watch,” says Park, the Go commentator. Such matches are too complex for fans to follow, too flawless to be thrilling, he says.
Players can mimic AI’s opening moves, but in the middle game—where the board branches into too many possibilities to memorize—their own judgment takes over. Fans revel in watching players make mistakes and mount comebacks, exuding personality in every stone on the board. Shin's playing style is combative but marked by machinelike poise. Kim deftly navigates the most chaotic positions on the board.
"In Go, every move is a choice you make, and your opponent responds with a choice of their own," says Kim Dae-hui, 27, a Go fan and amateur player. “Watching that process unfold is fun.”
With fans like Kim still watching, Shin finds meaning in his game. “I can play a kind of Go that tells a story that only a human can,” he says.
After his retirement, Lee searched for a new job where he could have an edge as a human. He started making board games, giving speeches, and teaching students at a university. “I’m looking for a new domain that I can enjoy and excel at,” he says.
But lately, he feels more hopeful for the game he left behind. “It’s every Go player’s dream to play a masterpiece game,” he says—a game of technical brilliance, with no mistakes, fought to a razor’s edge between evenly matched players. “It’s like a mirage,” Lee says, chuckling. “Maybe AI can help us play a masterpiece.”
Shin hopes he can do that. To Shin, AI is a teacher, a companion, and a North Star. “I may be one of the strongest human players, but with AI around, I can’t be so arrogant,” he says. “AI gives me a reason to keep improving.”
Deep Dive
Artificial intelligence
A “QuitGPT” campaign is urging people to cancel their ChatGPT subscriptions
Backlash against ICE is fueling a broader movement against AI companies’ ties to President Trump.
Moltbook was peak AI theater
The viral social network for bots reveals more about our own current mania for AI as it does about the future of agents.
Meet the new biologists treating LLMs like aliens
By studying large language models as if they were living things instead of computer programs, scientists are discovering some of their secrets for the first time.
What’s next for AI in 2026
Our AI writers make their big bets for the coming year—here are five hot trends to watch.
Stay connected
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.