OpenAI is giving artificial intelligence researchers a new way to test and evaluate their research. The organization has announced Universe, a software platform designed to train and measure the general intelligence of AI across games, websites and applications.
“Universe allows an AI agent to use a computer like a human does: by looking at screen pixels and operating a virtual keyboard and mouse,” the OpenAI team wrote in a blog post. “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.”
Researchers will also be able to turn their programs into Gym environments. (Gym is OpenAI’s recently launched toolkit for reinforcement learning.)
The new platform already has support and permission to access games and apps from EA, Microsoft Studios, Valve and Wolfram. OpenAI plans to integrate more solutions in the future.
While AI has seen major advancements in the past couple of years, OpenAI explains AI agents cannot bring their experiences to new tasks yet, which is why they created Universe. “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 toward general intelligence,” the team wrote.
“If we are to make progress toward generally intelligent agents, we must allow them to experience a wide repertoire of tasks so they can develop world knowledge and problem-solving strategies that can be efficiently reused in a new task.”
Environments currently include Atari games, Flash games, extracting rewards, browser tasks, browser interactions, and real-world browser tasks.
OpenAI plans to release a “transfer learning benchmark” to help researchers determine the progress on their experiments.