The Center for Human-Compatible AI (CHAI) is an AI research center at the University of California, Berkeley, founded in 2016 by Stuart Russell. Its goal is to develop the theoretical foundations for AI systems that remain compatible with human values.
Its main research topics include assistance games, in which AI cooperates with people while remaining uncertain about their preferences, value alignment, and corrigibility, the property that allows humans to correct or shut down an AI system. Researchers from Berkeley and several other universities participate in the center.