A New Artificial Intelligence Makes Mistakes—on Purpose
A chess program that learns from human error might be better at working with people or negotiating with them. It took about 50 years for computers to eviscerate humans in the venerable game of chess.…