Human Saga

AlphaGo defeats Lee Sedol, a milestone of the deep-learning revolution

9–15 March 2016·Science, technology & invention·South Korea·
9–15 March 2016Date
Science, technology & inventionCategory
0Empires active
Date
9–15 March 2016
Category
Science, technology & invention
Region
South Korea
Era
Contemporary
Significance

Between 9 and 15 March 2016 in Seoul, South Korea, AlphaGo, a program developed by the London-based DeepMind, defeated the world-class player Lee Sedol four games to one in the ancient board game Go.

Go had long resisted computer mastery because the number of possible positions vastly exceeds those of chess, making brute-force search impossible. AlphaGo combined deep neural networks with reinforcement learning and tree search, learning from human games and millions of games of self-play; its unconventional 'Move 37' in game two stunned experts. Watched by hundreds of millions worldwide, the victory was a landmark demonstration of the deep-learning revolution then transforming artificial intelligence. That revolution had been catalysed by the 2012 'AlexNet' breakthrough in image recognition and would be supercharged by the 2017 invention of the transformer neural-network architecture. Together these advances drove rapid progress in machine translation, speech and image recognition, and ultimately the large language models behind the generative-AI systems that reshaped technology in the 2020s.

See the world map in 9–15 March 2016 →More science, technology & inventionThe Contemporary era

One piece of history in your inbox, weekly

New empires, events and features as they land — one good email a week. No spam, unsubscribe anytime.

Related science, technology & invention events

Further reading

The wider story

A Little History of Science

William Bynum

Warm, jargon-free tour of discovery from Babylonian stargazers to the genome.

Get it on Amazon →

As an Amazon Associate, Human Saga earns from qualifying purchases.