AlphaGo (A): Birth of a New Intelligence

This case, the first of a three-part series, traces DeepMind's evolution from its 2010 founding through its acquisition by Google in 2014. Often referred to as the "Apollo project" of artificial intelligence, DeepMind used games as a testing ground to develop AI systems capable of learning complex tasks without explicit programming. Their greatest challenge was mastering Go—an ancient game whose complexity (with more possible board positions than atoms in the universe) and reliance on human intuition had made it the holy grail of AI research.
In March 2016, DeepMind's AlphaGo defeated Go world champion Lee Sedol 4-1 in a match watched by over 200 million people. During this victory, AlphaGo played the now-famous "Move 37"—a play so creative and counterintuitive that it forced experts to reevaluate their understanding of both Go and AI. While AlphaGo's victory demonstrated AI's ability to master tasks requiring human-like intuition and strategic foresight, Sedol's win in Game 4 exposed vulnerabilities in the system. As CEO Demis Hassabis and his team prepared for a 2017 match against the world's top-ranked Go player, they had to decide how to strengthen AlphaGo's architecture while preserving the creative gameplay that had captured the world's imagination.