Google develops AI that learns by itself

Google-owned artificial intelligence (AI) firm DeepMind has developed a new software program that learned by itself how to play the ancient Chinese strategy game Go in just three days.

Called AlphaGo Zero - after the AlphaGo programme that defeated a champion of the game in 2015 - the AI software simulated matches against itself to learn from its mistakes, says UK-based DeepMind.

The AI is so powerful that, having been told only the rules of Go, it derived thousands of years of human knowledge of the game before inventing better moves of its own.

Subscribe to The Week

Escape your echo chamber. Get the facts behind the news, plus analysis from multiple perspectives.

SUBSCRIBE & SAVE

Sign up for The Week's Free Newsletters

From our morning news briefing to a weekly Good News Newsletter, get the best of The Week delivered directly to your inbox.

From our morning news briefing to a weekly Good News Newsletter, get the best of The Week delivered directly to your inbox.

Sign up

And the software could be reprogrammed to discover information in other fields, such as “drug discovery” and “particle physics”, says The Verge.

DeepMind’s chief executive, Demis Hassabis, told BBC News: “We’re quite excited because we think this is now good enough to make some real progress on some real problems, even though we’re obviously a long way from full AI”.

Project leader David Silver told the news site that he hopes AI programs similar to AlphaGo Zero can be used for “advancing the frontiers of science and medicine”.

But while such technology could solve some serious global problems, humanity needs to “keep a close eye on the ethical dilemmas” that come with a machine that can think for itself, warns BBC technology reporter Rory Cellan-Jones.

However, there are “ few signs that AlphaGo Zero and its ilk will either steal our jobs or threaten to make humanity obsolete”, for now, at least, he adds.