• Research Active offline policy selection
• Research GopherCite: Teaching language models to support answers with verified quotes
• Research Learning Robust Real-Time Cultural Transmission without Human Data
• Science Accelerating fusion science through learned plasma control
• Blog MuZero’s first step from research into the real world
• Research Red Teaming Language Models with Language Models
• Research Spurious normativity enhances learning of compliance and enforcement behavior in artificial agents
• Research On the Expressivity of Markov Reward
• Company Real-world challenges for AGI
• Research Stacking our way to more general robots