The newest AlphaGo mastered the game with no human input | Science News

SUPPORT SCIENCE NEWS

Science News is a nonprofit.

Help us keep you informed.


News

The newest AlphaGo mastered the game with no human input

The computer program even devised new strategies previously unknown to human players

By
1:00pm, October 18, 2017
Go stones on a Go board

GAME CHANGER  The first completely self-taught AlphaGo program tops all of its machine-learning predecessors, no contest.

AlphaGo just leveled up.

The latest version of the computer program, dubbed AlphaGo Zero, is the first to master Go, a notoriously complex Chinese board game, without human guidance. Its predecessor — dubbed AlphaGo Lee when it became the first computer program with artificial intelligence, or AI, to defeat a human world champion Go player (SN Online: 3/15/16) — had to study millions of examples of human expert moves before playing practice games against itself. AlphaGo Zero trained solely through self play, starting with completely random moves. After a few days’ practice, AlphaGo Zero trounced AlphaGo Lee 100 games to none, researchers report in the Oct. 19 Nature.  

“The results are stunning,” says Jonathan Schaeffer, a computer

This article is only available to Science News subscribers. Already a subscriber? Log in now. Or subscribe today for full access.

Get Science News headlines by e-mail.

More from Science News

From the Nature Index Paid Content