Artificial intelligence learns to compromise. Image by pixabay.com
Could Artificial intelligence strike a bargain better than humans?

Artificial intelligence has made major advances and is now able to beat humans in Chess and also at Go. However, could we teach AI to also co-operate and compromise? Researchers at Brigham Young University have developed an algorithm to do just that. Using an algorithm S# they programmed machines for a two-player game to see how they would cooperate with humans. The machines were programmed to use common phrases like, 'Sweet. We are getting rich', or 'In your face!' and the human counterparts in the game at times couldn't distinguish between a machine and a human. The use of cheap talk by the machines surprisingly increased the level of cooperation. This shows that using simple algorithms machines can be trained to cooperate with humans which might be at times better than humans themselves.

Read the full story: www.neurosciencenews.com
Scientific publication: www.nature.com