Those among us who fear that we’ve already passed the point of no return when it comes to artificial intelligencebecoming self-aware and plotting to murder the human race will likely cite A.I. research company DeepMind’s latest experiment as further proof of that notion. Using Id Software’s Quake III Arena, DeepMind has managed to train artificial players to be even more effective than their human counterparts.
The challenge for DeepMind was not to see if its A.I. agents could defeat human players in battle, but rather if they could work together on procedurally generated levels to complete an objective — in this case, capture the flag.
Because the levels’ structure changes each time they play, the agents are unable to simply memorize locations in order to make it to the flag. This forced them to actually learn the strategies needed to win in a similar manner to how human players might improve at the game. The agents even used an emulated game controller to control their characters in Quake III Arena.
“The agents are never told anything about the rules of the game, yet learn about fundamental game concepts and effectively develop an intuition for capture the flag,” DeepMind said in a blog post. “In fact, we can find particular neurons that code directly for some of the most important game states, such as a neuron that activates when the agent’s flag is taken, or a neuron that activates when an agent’s team is holding the flag.”
As the A.I. agents won matches, the behavior they used to do so was reinforced and these agents were paired up with human players to impressive results. Eventually, DeepMind held a Quake III Arena tournament, and not only did these A.I. agents have a higher success rate than humans, they were also rated as “more collaborative” in a post-tournament survey. Unfortunately, just like with so many children on Xbox Live, the A.I. agents also learned to “camp.” Maybe it is a legitimate strategy.
You can watch a full deep-dive video on the Quake III Arena project above. Visual settings have been greatly reduced, as the A.I. don’t actually care how their game looks.