Google-owned UK artificial intelligence firm DeepMind developed an AI system that can play all 57 Atari 2600 video games and beat human averages, according to a research paper published this week.
The AI, called Agent57, is the first that can outperform standard human benchmarks for the games, the paper said.
"Games are an excellent testing ground for building adaptive algorithms: they provide a rich suite of tasks which players must develop sophisticated behavioural strategies to master, but they also provide an easy progress metric -- game score -- to optimise against," the researchers wrote in a blog post about the research. "The ultimate goal is not to develop systems that excel at games, but rather to use games as a stepping stone for developing systems that learn to excel at a broad set of challenges."
Agent57 builds on DeepMind's previous AI work: In March 2016, its AlphaGo machine learning system beat world champion Lee Sedol at the strategic game Go.
