DeepMind has developed an AI algorithm called MuZero that can master Go, chess and video games without being given the playing rules, attaining “superhuman performance”.
Researchers at DeepMind calls it “a significant step forward in the pursuit of general-purpose algorithms”.
The MuZero algorithm follows on from AlphaGo, the first artificial intelligence (AI) program introduced by DeepMind in 2016 to defeat humans at the ancient game of Go.
Two years later, its successor – AlphaZero – learned from scratch to master Go, chess and shogi.
Instead of trying to model the entire environment, MuZero focuses only on the most important aspects of the environment that are important to the agent’s decision-making process.
“By combining this model with AlphaZero’s powerful lookahead tree search, MuZero set a new state of the art result on the Atari benchmark, while simultaneously matching the performance of AlphaZero in the classic planning challenges of Go, chess and shogi. In doing so, MuZero demonstrates a significant leap forward in the capabilities of reinforcement learning algorithms,” explained an official blog post.