Human against machine
10¹⁷⁰possible positions on the Go board
In March 2016 DeepMind's program AlphaGo played five games in Seoul against Lee Sedol, winner of 18 world titles. Go is considered vastly more complex than chess. More than 200 million people around the world watched.
Your guess
How likely was it that a human would play AlphaGo's famous move 37?
The human's answer
AlphaGo won the first three games. In the fourth, Lee, playing white, made an unexpected move 78. The professional Gu Li called it a "divine move"; it turned the game around. It remained Lee's only win.
What came after
4:1AlphaGo won the match
Many top players described AlphaGo's moves as questionable at first but sensible in hindsight. The program had first learned from about 30 million moves from human games – and ended up playing moves that many pros saw as entirely its own.
