The ancient Chinese game of Go is one of the last games where the best human players can still beat the best artificial intelligence players. Last year, the Facebook AI Research team started creating an AI that can learn to play Go.

Scientists have been trying to teach computers to win at Go for 20 years. We're getting close, and in the past six months we've built an AI that can make moves in as fast as 0.1 seconds and still be as good as previous systems that took years to build. Our AI combines a search-based approach that models every possible move as the game progresses along with a pattern matching system built by our computer vision team.

The researcher who works on this, Yuandong Tian, sits about 20 feet from my desk. I love having our AI team right near me so I can learn from what they're working on.

You can learn more about this research here: http://arxiv.org/abs/1511.06410

[He also posted a video: https://www.facebook.com/zuck/videos/10102619979032811/]

We've issued the following Press Release today (27th January 2016):

A computer program developed by Google DeepMind (AlphaGo) to play the Oriental game of Go has beaten the three-times European Go Champion and Chinese professional Fan Hui (shown on the right in the photo, courtesy of Google DeepMind). This is the first time that a Go-professional has lost such a match, and not only that, by a clean sweep in all 5 games. This signifies a major step forward in one of the great challenges in the development of artificial intelligence - that of game-playing.

These findings were reported in a peer-reviewed study published in the scientific journal Nature: Silver D. et al. Mastering the game of Go with deep neural networks and tree search.

Andrew Simons (shown right playing Des Cann) won all his games to win the Maidenhead-Hitachi Tournament, held as normal at the Hitachi Europe HQ, thanks to Hitachi.

Also winning three games were Roger Huyshe (3k Shropshire), Malcolm Hagan (5k Winchester), Stephen Bailey (7k Arundel) and Robert Scantlebury (9k Sheffield). Paul Barnard (2k Swindon) won 2.5. All players on two wins won a prize each too, with Hitachi maintaining their generous Go-support with a fine array of prizes.

Best teams were Arundel (Steve, Malcolm, Pauline and Casey) on 9/12 and "No Go" (Roella, Edmund, Charlotte and Zaki) on 8/12. The 13x13 prizes went to Edmund Smith (8k Milton School) with 3/4 and Lily Danson (15k Cheadle Hulme School) with 4/7.

Results

The UK Youth GO team have finished a respectable sixth out of the twelve teams competing in the European online Youth Go Team Tournament. This year the draw was an accelerated Swiss which meant the top teams were paired against each other from an early stage. This gave our team more chances to play more evenly matched opponents. We picked up wins against Poland (3-2) and Italy, Switzerland and Austria (4-1) and lost to Czechia (2-3) and Croatia, Slovenia and Serbia (1-3).

As last year we struggled most on the U20 board where we do not yet have any dan players. However our U12s continue to excel in this event - we picked up wins in 6 or the 7 games played in this age group with two of our U12 players also playing up an age group on occasion.

The nominations for our Annual Awards are now closed and can be seen here: http://www.britgo.org/awards/2016

Voting is now open to our Association Members (you need to login first) and will close one week before our AGM, on 24th March.

The 42nd London Open was dominated by Sai Sun (5d) from Beijing who won three titles. She won the Open title and £120, with six wins out of seven, won the Lightning and also the Pair Go with Xinchi Gong. In all 76 players took part in the event, held as usual at ISH in London.



On hand to do game analysis and lectures was the professional from Romania, Catalin Taranu (5p). He lectured on the opening and analysed the game between Andrew and Sai.

Congratulations to the Cambridge A team (Lou Yuxiang, Paul Smith, Atta Chui and James Wood) for winning the seventh season of the online league. In second place were Edinburgh 1. Winning division 2 and promotion to division 1 were UK Youth.

The winning teams received prizes consisting of places in Guo Juan's Group Classes and subscriptions to the Internet Lectures and Training system. See the Internet Go School site for more details.

League Details

Despite being very near to Christmas and at a new venue near Edinburgh Castle, St Columba's by the Castle Episcopal Church Hall, 16 players managed to take part.

Tied at the top on three wins out of four and equal SOS were "Henry" Hongyi Chen (2d) from Glasgow and Polish player Jakub Ziomko (1d) from Aberdeen.

Two further prizes for three wins were awarded to Ben Lloyd (3k Edinburgh) and Ron Bell (5k Borders).

Results

The UK Youth team had an exciting match today against the combined team of Croatia, Serbia and Slovenia. First to finish was Aidan on board 4 who played a solid game and claimed a straight forward win. We then lost on board 1 where Elom had a tough game against a 3 dan opponent. With the score 1-1 we needed a win on one of the two U16 boards. Both Charlotte and Edmund had exciting games which looked to be going their way after the opening. Charlotte had killed a large group and Edmund had mapped out a massive moyo. Unfortunately their opponents fought back strongly and we could not hold onto the advantage in either game leaving us 1-3 down with board 5 to play. Rather oddly neither board 5 managed to turn up online. Wenzhou for the UK had connection problems preventing him joining the match. The other team's player turned up 33 minutes late-technically also a default.

The Facebook AI Research group have submitted this paper for publication at ICLR 2016 (a conference in Puerto Rico in May). So the rumours of their developing a Go-playing program are correct!

As they say - it's not up to the best commercial programs yet, but it's probably around 4 Dan ...

Better Computer Go Player with Neural Network and Long-term Prediction.

Play it on KGS - darkforest and darkfores1.