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.

source https://www.wired.com/story/new-artificial-intelligence-mistakes-purpose-chess