Google DeepMindFrom the builder
From Atari to EVE Online: Building on 15 Years of AI Research in Games
Google DeepMind looks back on games research since its 2010 founding, from the DQN agent that learned 49 Atari games and AlphaGo's win over Lee Sedol to SIMA 2, a Gemini-powered generalist agent that follows natural language instructions. It also describes its partnership with Fenris Creations, maker of EVE Online, whose persistent multiplayer universe of more than two decades suits research on continual learning, memory, long-horizon planning and multi-agent dynamics; the work has already produced a Gemini-based guidance system for new players, and further research starts in an offline instance kept apart from live players.
Games have always been a mirror for intelligence.
Saved . A shortened excerpt from the official blog; the original publisher retains authorship.