As explained on OpenAI's blog, Universe makes it possible for an AI agent to interface with a computer like a human does. Just like a real human, Universe looks at pixel output and operates a (virtual) keyboard and mouse.
Today's public release features thousands of environments that can be used to train AI agents, including numerous Flash games, browser tasks, and games like Portal, Fable, Red Alert 2, Sid Meier's Alpha Centauri, etc. The end-goal of this particular project is to develop a single AI agent that can apply past experience in new environments.
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.Below is an example of three AI agents playing Flash-based racing games, head over to the OpenAI blog for a lot more details about how this works and what the future looks like.