Publications
Publications
- November 2024
- HBS Case Collection
AlphaGo (B): Birth of a New Intelligence
By: Shikhar Ghosh and Shweta Bagai
Abstract
This case, the second in a three-part series, explores DeepMind's evolution from developing game-specific AI to more generalized learning systems. Following AlphaGo's 2017 victory over the Go world champion, DeepMind introduced two revolutionary systems that eliminated the need for human data: AlphaGo Zero, which mastered Go through pure self-learning, and AlphaZero, which achieved superhuman performance in chess, Go, and shogi within hours using the same self-play principles. While AlphaGo had relied on studying human games, these new systems demonstrated that AI could develop superhuman abilities through self-learning alone—surpassing their predecessors in both performance and efficiency.
The breakthrough had implications that extended beyond gaming. China declared AlphaGo's victory a "Sputnik moment" and announced an ambitious plan to lead global AI development by 2030. Russia's president proclaimed that the nation dominating AI would rule the world. As DeepMind contemplated the next frontiers for their technology, CEO Demis Hassabis faced critical questions: Could these self-learning principles be adapted to tackle complex real-world problems in science and other domains? How would nations respond as AI's potential expanded beyond games into strategic sectors of the economy?
Keywords
Citation
Ghosh, Shikhar, and Shweta Bagai. "AlphaGo (B): Birth of a New Intelligence." Harvard Business School Supplement 825-074, November 2024.