Artificial intelligence is getting closer to taking on and beating humans even in highly complex and unstructured games. DeepMind announced that its AI agents exceed human level gameplay in Quake III. This is an ever-changing 3D environment in which one captures the flag. The scientists who prepped the play had their AI program compete each time in a new configuration as it learned. Hence, it could not solve the problem in one space alone, but had to learn how to play in all of them. After about 200,000 training games, it passed the strong human level of play. Along with AI’s success at Go and Chess, it is one more leap in computers’ abilities to transcend humans. Some are terrified that computers will take over some day. Others say it is highly unlikely. While I’m in the latter camp, the power of AI is becoming a PR problem for developers and companies using it. They need to communicate its limitations and not just its abilities. Otherwise, it could well become a political issue in which AI is regulated.