backtop


Print

Google DeepMind says it has made another big advance in artificial intelligence by getting a machine to master the Chinese game of Go without human players help.

The tech giant’s AI division devised the AlphaGo program, and had beaten two of the world’s best players.

 

 

It started by learning from many games played by humans, but the new AlphaGo began with a blank Go board and no data apart from the rules, and then played itself.

 

It is reported, that within 72 hours it was good enough to beat the original program by 100 games to zero.

 

Demis Hassabis chief executive of DeepMind Said, the system could now have more general applications in scientific research.

 

He told BBC,” even though we are obviously a long way from full AI, we think this is now good enough to make some real progress on some real problems.”

 

Also, it is reported that the London-based artificial intelligence company’s software defeated leading South Korean Go player Lee Se-dol by four games to one last year.

 

Where the game has more possible moves than are atoms in the universe, this was a big victory for machine over man, that came earlier than many in the AI world expected. 

 

David Silver, who led the effort, says the team took a very different approach with AlphaGo Zero. He explained, “ the new version starts from a neural network that knows nothing at all about the game of Go.”

 

 

Go was thought to date back to ancient China many years ago. Go uses white and black stones on a grid, similar to chess with two different color armies that players use to gain the upper hand by surrounding their opponents.  

 

Go’s rules are simpler than chess rules but the number of moves in chess is about 20 compared to Go that has a choice of 200 moves at most points in the game. 

 

It can be very difficult to determine who is winning, and this turned out to be a more efficient way of addressing the problem.

 

The AIphaGo Zero developed techniques became so interesting and powerful that many of the team has moved on to new projects trying to take this technique to new areas.  Demis Hassabis mentioned drug design and the discovery of new materials as areas of interest.





"Can anyone tell me what MobileMe is supposed to do?... So why the f*** doesn't it do that?" -- Steve Jobs






Most Popular Articles







botimage
Copyright 2018 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki