In March 2016, the world held its breath watching a match of the century. It was the game between AlphaGo, the artificial intelligence Go program from Google DeepMind, and Lee Sedol 9-dan, the top human Go player. At the time, even experts predicted it would be difficult for AI to beat a human, but the result was AlphaGo’s victory with 4 wins and 1 loss. This shocking event became a monumental milestone, signaling the dawn of the artificial intelligence era.
Exactly 10 years later, in 2026, we are living amidst the questions and changes AlphaGo brought forth. Shall we frankly discuss how the events on the Go board back then impacted the future of humanity beyond a simple game, and how far AI has come today?
The Match of the Century: AlphaGo and Lee Sedol 9-dan
Honestly, everyone was surprised watching that match, weren’t they? The very idea of artificial intelligence beating a human in Go was hard to imagine. AlphaGo had already won against Fan Hui 2-dan in 2015, but the match against Lee Sedol 9-dan truly captivated the attention of people worldwide. In particular, the 4th game, where Lee Sedol 9-dan achieved his sole victory, is still talked about today for its ‘divine move’.
This match went beyond the rules and strategies of the game of Go, prompting deep reflection on the potential of artificial intelligence. Many believed that AI could not mimic human intuition and creativity, but AlphaGo made unexpected moves, shaking up the conventional wisdom of the Go world.
- AlphaGo’s Groundbreaking Victory: In March 2016, it defeated Lee Sedol 9-dan 4-1, impressing the world with the capabilities of artificial intelligence.
- Lee Sedol 9-dan’s ‘Divine Move’: Lee Sedol 9-dan’s 78th move in the 4th game is regarded as a move that induced an error in AlphaGo, remembered as a masterpiece demonstrating human creativity.
- Paradigm Shift in the Go World: After AlphaGo’s appearance, professional Go players began to research new joseki and fuseki through AI, completely changing the way Go is studied.
AlphaGo’s Evolution and the Advancement of AI Technology
AlphaGo was not just a project for Go. The DeepMind development team developed this technology with the goal of a single artificial intelligence that could be applied to all IT services. Even after the match with Lee Sedol 9-dan, AlphaGo continued to evolve. ‘AlphaGo Zero’ emerged, becoming even more powerful by learning purely from self-play without human game records, and then ‘AlphaZero’, a general-purpose AI that mastered not only Go but also chess, shogi, and all two-player perfect information games.
The core of this evolution lies in technologies like deep learning and reinforcement learning. AlphaGo improved its skill by learning from vast amounts of data and finding optimal moves through self-play. Initially, it used GPUs, but later, Google’s self-developed TPUs (Tensor Processing Units) enabled more efficient computation.
- AlphaGo’s Core Technology: Combines trained deep neural networks (DNN) with Monte Carlo tree search (MCTS) to find optimal moves.
- Evolved Learning Methods: Initially, it used both supervised learning and reinforcement learning, but subsequent models like AlphaGo Master and Zero improved their skill solely through reinforcement learning without supervised learning.
- Diverse Hardware Utilization: Initially used CPUs and GPUs, and later utilized TPUs specialized for deep learning to maximize computational efficiency.
2026: The Significance of 10 Years After AlphaGo
In 2026, the 10th anniversary of the AlphaGo vs. Lee Sedol 9-dan match, artificial intelligence has become such a natural part of our lives. In the Go world, AI acts as a ‘teacher’, proposing new fuseki and strategies, and Lee Sedol 9-dan even participated in an experiment to design ‘future Go’ with AI again after 10 years. Lee Sedol 9-dan now goes so far as to say, “AI is simply a god.”
AlphaGo has retired from Go, but its technology has expanded into various fields such as drug discovery, climate change prediction, autonomous vehicles, and medical diagnosis. In particular, in 2026, expectations for ‘physical AI’ with physical forms like autonomous cars and robots are growing. The possibilities demonstrated by AlphaGo continue to inspire the journey towards artificial general intelligence (AGI). Isn’t it truly amazing that the innovation that began on a Go board is now contributing to solving complex problems in the real world?
AlphaGo was more than just a program that played Go well; it posed fundamental questions about how far artificial intelligence can develop and how humans and AI can coexist. 10 years later, we are in the process of finding answers to those questions, and it seems inevitable that we are even more excited about the future that AI will create.