This week, the computer that learnt to play Go.

Everyone in deep mind was crowded into a room, brimming with excitement, watching AlphaGo's

internal evaluation go up and up and up as it started to believe it was winning.

And researchers watch 100,000 worms live and die.

I think this is possibly the single best lifespan experiment ever conducted.

Plus, the Lost Library of Tudor Scholar John D.

This is the Nature podcast for January the 28th, 2016.

I'm Adam Levy.

And I'm Kerry Smith.

Fun and games with artificial intelligence first this week.

A computer has learned to play the ancient game of Go.

Lizzie Gibney reports.

Computers beat humans at lots of things.

Crunching numbers, remembering long lists, tasks that require a lot of processing power or memory, but not a lot of creativity. In looking for bigger challenges, developers of artificial intelligence have often turned to games. Games need long-term planning, predictive power and cunning, in addition to all the processing power that computers have in spades. Chess was perhaps the most famous one to fall.

Chess Grandmaster Gary Kasparov lost a deep blue way back in 1997.

David Silver, from Google-owned company Deep Mind, explains.

Most board games were actually relatively straightforward for AIs to actually defeat humans

because they could basically use brute force search.

Brute Force search, basically combing through all the moves,

imagining the game several turns ahead and choosing the best play.

So say your AI is playing chess.

Black chooses his move and White considers all the possible follow-ups and black considers all the follow-ups to that.

...