Researchers at the DeepMind artificial intelligence (AI) laboratory have developed automated "agents" that display humanlike behavior when playing in capture-the-flag mode in the Quake III multiplayer computer game, exhibiting coordinated teamwork against or in collaboration with human participants.
The agents learned skills like racing toward the opponent's home base when a teammate was about to capture a flag from approximately four years' worth of accumulated gameplay.
Many researchers hope such breakthroughs will lead to more effective automated systems with real-world applications, like driverless cars that navigate through traffic.
Many AI experts are skeptical the technology could be used to address real-world problems. Georgia Institute of Technology's Mark Riedl says DeepMind's agents are not truly collaborating, but responding to what is happening in the game.
From The New York Times
View Full Article - May Require Paid Subscription
Abstracts Copyright © 2019 SmithBucklin, Washington, DC, USA