NPR

Computer Learns To Play Go At Superhuman Levels 'Without Human Knowledge'

Researchers say they have constructed an AI program that can teach itself to play the ancient strategy game at a level far beyond humans.
According to the researchers, there are 10 to the power of 170 possible board configurations in Go — more than the number of atoms in the known universe.

A year after a computer beat a human world champion in the ancient strategy game of Go, researchers say they have now constructed an even stronger version of the program — one that can teach itself without the benefit of human knowledge.

The program, known as AlphaGo Zero, became a Go master in just three days by playing 4.9 million games against itself in quick succession.

"In a short space of time, AlphaGo Zero has understood all of

You’re reading a preview, subscribe to read more.

More from NPR

NPR2 min read
The Louvre Museum Looks To Rehouse The 'Mona Lisa' In Its Own Room — Underground
Louvre Director Laurence des Cars said her institution is looking at upgrading both the visitor experience surrounding the iconic painting as well as the museum overall.
NPR4 min read
Got Brothers Or Sisters? Warm Sibling Bonds Help Booster Happiness As You Age
Researchers have found that a warm, close bond with a sibling in early adult life is predictive of good emotional health later in life, with less loneliness, anxiety and depression.
NPR3 min read
What's Making Us Happy: A Guide To Your Weekend Viewing, Listening And Gaming
Each week, Pop Culture Happy Hour guests and hosts share what's bringing them joy. This week: the show Baby Reindeer, Chappell Roan's Tiny Desk concert, a podcast about MTV, and retro gaming.

Related Books & Audiobooks