Nature | Blog

The Go Files: AI computer wins first match against master Go player

Nature reports from a battle of man vs machine over the Go board.

Updated:

Article tools

Rights & Permissions

Google DeepMind

Lee Sedol (right) analyses his game after resigning in his first match against AlphaGo.

Tanguy Chouard, an editor with Nature, saw Google-DeepMind’s AI system AlphaGo defeat a human professional for the first time last year at the ancient board game Go. This week, he is watching top professional Lee Sedol take on AlphaGo, in Seoul, for a $1 million prize.

As I’m sure you’ve heard, one of humanity’s best Go players, Lee Sedol, just lost his first game (replay below) in a best-of-five series against AlphaGo, the artificial intelligence system created by Google DeepMind. “I’m in shock,” Lee said at the post-match press conference.

I was surprised too by my emotional reaction to the result. Before I saw the game, which I watched alongside DeepMind’s CEO Demis Hassabis, and David Silver, one of the lead researchers on AlphaGo, I didn’t know who I wanted to win.

Read more of Tanguy's blog at The Go Files.

Now – to my surprise – I find myself rooting for the very personable and human Lee Sedol.

For me, the key moment came when I saw Hassabis passing his iPhone to other Google executives in our VIP room, some three hours into the game. From their smiles, you knew straight away that they were pretty sure they were winning – although the experts providing the live public commentary on the match that was broadcast to our room weren’t clear on the matter, and remained confused up to the end of the game just before Lee resigned. (I'm told that other high-level commentators did see the writing on the wall, however).

Hassabis’s certainty came from Google’s technical team, who pore over AlphaGo’s evaluation of its position, information that isn’t publicly available. I’d been asking Silver how AlphaGo saw the game going, and he’d already whispered back: “It’s looking good”.

And I realised I had a lump in my throat. From that point on, it was crushing for me to watch Lee’s struggle.

Towards the end of the match, Michael Redmond, an American commentator who is the only westerner to reach the top rank of 9 dan pro, said the game was still “very close”. But Hassabis was frowning and shaking his head – he knew that AlphaGo was definitely winning. And then Lee resigned, three and a half hours in.

At the post-game press conference, Lee said he felt he’d made a mistake early in the game, and had to drag that burden throughout the match.  We’d certainly seen Lee looking tense early on: Redmond had already told us – and the hundreds of journalists watching in two overflow rooms – that AlphaGo was playing more aggressively, in contrast to its relatively peaceful games last year against Fan Hui. 

"AlphaGo played like a human professional player, but with the emotional element carved out,” said one Korean commentator, Kim Sung-ryong.  He said that all top-level pro players were in shock. Silver said that – judging from the statistics he’d seen when sitting in Google’s technical room – “Lee Sedol pushed AlphaGo to its limits”.

Does this mean that AI has finally defeated humanity at what may be the most complex board game we've ever invented? Not quite. There are still four games to go; tomorrow (10 March, livestream below) we may see a different kind of match, as AlphaGo will play first, with the black stones, rather than second, with the white ones.

Lee said he wouldn’t be affected psychologically by the loss. “This game is not going to affect how I play in the next games,” he said. As Redmond told me after the game: "he's a very tough guy".

Can't wait to see him fight back!

Previous entry: Champion preps for $1 million machine match

Journal name:
Nature
DOI:
doi:10.1038/nature.2016.19544

Updates

Updated:

This blog was updated to clarify that some commentators — although not the ones Tanguy heard during the match — recognized at around the same time as the DeepMind team did that Lee Sedol was behind in the game.

For the best commenting experience, please login or register as a user and agree to our Community Guidelines. You will be re-directed back to this page where you will see comments updating in real-time and have the ability to recommend comments to other users.

Comments for this thread are now closed.

Comments

2 comments Subscribe to comments

  1. Avatar for Louis Oldershaw
    Louis Oldershaw
    Unless one subscribes to the "Ghost In Machine" version of human thought, this step forward for AI is not the least bit surprising. When Deep Blue beat Gary Kasparov at chess the writing was clearly on the wall, so to speak. The human brain and its thought output may be exceedingly subtle and complex, but the brain is only a very large number of neuronal parts and causal physical interactions. It may take a long time before computers can mimic everything a human can do, but to assume that computers cannot do so in principle is a very dubious assumption -one without much scientific merit. I cannot wait for a really advanced computer system to process all available, well-tested, observational physics/astronomy data and then formulate a cosmological paradigm. I would bet dollars to donuts that it would do one hell of a lot better than our current crop of theoretical physicists.
  2. Avatar for David Britt
    David Britt
    This is a nice article, but it's simply untrue that the professionals observing didn't see the outcome until Lee's resignation. The Chinese commentary had white leading after the invasion on the right, more than 30 min from the end. The high level English commentary said the game was over after the bottom right was decided. The Korean commentary was understandably a little in denial. And Redmond didn't take time to fully count the board, since he was engaged in beginner-oriented commentary. Google didn't really have advance knowledge.

Refugee trauma

refugee-mental-health

The mental-health crisis among migrants

The refugees and migrants surging into Europe are suffering very high levels of psychiatric disorders. Researchers are struggling to help.

Newsletter

The best science news from Nature and beyond, direct to your inbox every day.

Deadly mutants?

genetic-mutants

A radical revision of human genetics

Why many ‘deadly’ gene mutations are turning out to be harmless.

Artificial intelligence

AI-blindspot

There is a blind spot in AI research

Fears about the future impacts of artificial intelligence are distracting researchers from the real risks of deployed systems.

Brain implant

paralysis

Pioneering brain implant restores paralysed man's sense of touch

More-advanced implants will be needed to restore full sensation throughout the body.

Galactic census

galaxies

Universe has ten times more galaxies than researchers thought

The new estimate could help astronomers better understand how galaxies form and grow.

Nature Podcast

new-pod-red

Listen

This week, refugee mental health, better neural nets, and changing attitudes to female genital cutting.

Science jobs from naturejobs