Google reveals how DeepMind AI learned to play Quake III Arena like a pro

A team of Google DeepMind researchers have trained AI agents to 'act independently, yet learn to interact and cooperate with other agents' by letting them play CTF in (what else?) Quake III Arena. ...