Google DeepMind Partners with Game Developers to Advance AI Research Through Gaming
Google DeepMind is collaborating with various game development studios to push the boundaries of artificial intelligence research using games as a testing ground. This initiative builds on DeepMind's 15-year history of leveraging games, from Atari to StarCraft, to achieve significant AI breakthroughs. The current focus is on developing 'general gaming agents' like SIMA (Scalable Instructable Multiworld Agent), which can understand natural language instructions and interact with game worlds using standard keyboard and mouse controls, without requiring API access or source code modifications. These agents are designed to 'see' what a player sees on screen and reason in real-time, aiming to achieve human-like play across complex 3D environments and video games such as No Man's Sky and Valheim. The partnerships with studios like Fenris Creations, Hello Games, and Coffee Stain Studios provide diverse and challenging environments for AI development. 
