Facebook Creates AI That Plays ‘Go’

Facebook says that it has designed an artificial intelligence (AI) software that has the ability to beat humans at a game of ‘Go’.

Facebook’s AI Research (FAIR) team has created a state-of-the-art learning tool that can distinguish between objects in a photo.

The team at Facebook has also taught the system to predict and plan. Scientists used a game of Go to demonstrate that the AI is as good as or even better than humans at planning.

“After a few months of playing, it’s already on par with the other AI-powered systems that have been published and it’s as good as a very strong human player. We’ve achieved this by combining the traditional search-based approach — modeling out each possible move as the game progresses — with a pattern-matching system built by our computer vision team,” Facebook said in a news release.

Go is incredibly tough for machines to crack, mostly due to the vast number of permutations required to make a move. As a general rule, computers play and win games by analyzing all possible outcomes of every move. In a game of Go, players move pieces along intersections, meaning that the number of possible moves is huge. Most players rely on intuition to make the right calls while playing the game. In essence, the research team is teaching the AI to look for visual data that will help it distinguish between objects in a given setting.

“Our AI research efforts — along with our work to develop radical new approaches to connectivity and our work to develop immersive new VR technologies — are a long-term endeavor,” the social network company said in a news release.

Share this: