View all newsletters
Receive our newsletter – data, insights and analysis delivered to you
  1. Technology
  2. Hardware
March 9, 2016updated 05 Sep 2016 7:21am

Google AI software beats Go world champion in man vs machine battle

News: DeepMind’s AI program won the first of five matches against South Korean Lee Se-dol.

By

In a historic machine vs man match, Google’s DeepMind artificial intelligence (AI) AlphaGo program has beaten the world champion of Chinese game Go.

The three hour match, held in Seoul, was the first of a series of five where Go pro Lee Se-dol is going head to head against Google’s AI software.

In January, the company revealed that AlphaGo had already beaten the European champion of the Chinese board game in a secret match hosted in London in October 2015.

Go is a 2,500-year-old board game which only requires a squared board and black and white stones.

The game starts with an empty board and each player has an unlimited supply of stones, according to the British Go Association.

The organisation explains that the main object of the game is to use the stones to form territories by surrounding vacant areas of the board. "It is also possible to capture your opponent’s stones by completely surrounding them."

Taking turns, players place one of their stones on a vacant point at each turn, with the person playing with black stones going first.

Content from our partners
Webinar - Top 3 Ways to Build Security into DevOps
Tech sector is making progress on diversity, but advances must accelerate
How to bolster finance functions and leverage tech to future-proof operational capabilities

Stones are placed on the intersections of the lines rather than in the squares and once played stones cannot be moved. However "they may be captured, in which case they are removed from the board, and kept by the capturing player as prisoners".

Go is seen as a more complex game than chess, as players have 200 move options at a time, compared to 20 in chess.

According to the BBC, Se-dol was defeated in the last 20 minutes of the game as AlphaGo moved its white stones.

He said: "I was very surprised because I did not think that I would lose the game. A mistake I made at the very beginning lasted until the very last."

Google’s DeepMind said that it has trained the neural networks on 30 million moves from games played by human experts, until it could predict the human move 57% of the time.

AlphaGo then learned to discover new strategies for itself, by playing thousands of games between its neural networks, and adjusting the connections using a trial-and-error process known as reinforcement learning. The system also makes extensive use of Google Cloud Platform.

Se-dol and Alpha go will go head to head four more times on March, 10, 12, 13 and 15. The winner of the five-game challenge will receive $1 million.

Alphabet chairman Eric Schmidt said: "Whenever you have a large number of people using something, we can probably use machine intelligence to make it more efficient."

Demis Hassabis, CEO of DeepMind took to Twitter to express his happiness following his AI program victory.

He said: "#AlphaGo WINS!!!! We landed it on the moon. So proud of the team!! Respect to the amazing Lee Sedol too

#AlphaGo has also become the sixth trendiest topic on the social network. The match has been broadcasted on South Korea’s TV and was shown live on 14 channels in China.

DeepMind is also live streaming the matches via its YouTube channel.

 

Websites in our network
NEWSLETTER Sign up Tick the boxes of the newsletters you would like to receive. Tech Monitor's research, insight and analysis examines the frontiers of digital transformation to help tech leaders navigate the future. Our Changelog newsletter delivers our best work to your inbox every week.
I consent to New Statesman Media Group collecting my details provided via this form in accordance with the Privacy Policy
SUBSCRIBED

THANK YOU