Universe

Universe allows an AI agent⁠(opens in a new window) to use a computer like a human does: by looking at screen pixels and operating a virtual keyboard and mouse. We must train AI systems on the full range of tasks we expect them to solve, and Universe lets us train a single agent on any task a human can complete with a computer.

The sample code above will start your AI playing the Dusk Drive Flash game. Your AI will be given frames like the above 60 times per second. You’ll need to have Docker and universe installed.

Our goal⁠ is to develop a single AI agent that can flexibly apply its past experience on Universe environments to quickly master unfamiliar, difficult environments, which would be a major step towards general intelligence. There are many ways to help⁠: giving us permission on your games, training agents across Universe tasks, (soon) integrating new games, or (soon) playing the games.

With support from EA, Microsoft Studios, Valve, Wolfram, and many others, we’ve already secured permission for Universe AI agents to freely access games and applications such as Portal⁠(opens in a new window), Fable Anniversary⁠(opens in a new window), World of Goo⁠(opens in a new window), RimWorld⁠(opens in a new window), Slime Rancher⁠(opens in a new window), Shovel Knight⁠(opens in a new window), SpaceChem⁠(opens in a new window), Wing Commander III⁠(opens in a new window), Command & Conquer: Red Alert 2⁠(opens in a new window), Syndicate⁠(opens in a new window), Magic Carpet⁠(opens in a new window), Mirror’s Edge⁠(opens in a new window), Sid Meier’s Alpha Centauri⁠(opens in a new window), and Wolfram Mathematica⁠(opens in a new window). We look forward to integrating these and many more⁠.

Background

Universe Infrastructure

Here are some important properties of our current implementation:

Environments

Atari games

Flash games

Browser tasks

Future integrations

Running an environment

Starting a runtime. You can boot your first runtime from the console as follows: