Elon Musk’s OpenAI And Google’s DeepMind Open Their AI Platforms To Researchers

Both Google and OpenAI announced plans to open-source their deep learning code Monday. Elon Musk’s OpenAI released Universe, a software platform for measuring and training an AI’s general intelligence across games, websites and other applications. Google’s parent Alphabet is putting its entire DeepMind Lab training environment codebase on GiftHub, making it easier for anyone to train their own AI systems.

DeepMind may have defeated a world champion at the difficult game Go, but to advance its learning further, Alphabet says that AI agents require more detailed environments to help with AI research. The company is opening-sourcing DeepMind Lab to programmers that want to use it.

“DeepMind Lab is a fully 3D game-like platform tailored for agent-based AI research,” Alphabet wrote in a blog. The agent floats around the environment, moving in different directions, with a virtual camera that can track around its body.

Alphabet says the first-person 3D environment should make AI smarter. “If you or I had grown up in a world that looked like Space Invaders or Pac-Man, it doesn't seem likely we would have achieved much general intelligence."

An example of the decision-making is shown in the video below, where an agent grabs apples instead of a bitter lemon in order to collect the ultimate reward, which is the melon.

It’s no surprise that OpenAI, run by Elon Musk, is releasing a similar platform called Universe at the same time. As with DeepMind Lab, the idea is to give researchers a way to test their agents. OpenAI’s mission is to release “a thousand environments including Flash games, browser tasks and games like slither.io and GTA.V,” the group wrote.

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.

OpenAI says “AlphaGo can easily defeat you at Go, but you can’t explain the rules of a different board game to it and expect it to play with you.” It’s allowing AI to run different type of tasks, “so they can develop world knowledge and problem solving strategies that can be efficiently reused in a new task.”

OpenAI has received permission from EA, Microsoft Studios, Valve and other companies to access games like Wing Commander III, Portal and Rimworld for deep learning.

The DeepMind Lab will go live on GitHub later this week and OpenAI’s Universe is now available.