Who is betting on what in AI, and against whom.
Researcher
Co-author of the standard AI textbook and founder of Berkeley's Center for Human-Compatible AI. Argues that the field should build systems that are uncertain about human objectives, a design constraint that cuts across every camp rather than a bet on one.
Known for Artificial Intelligence: A Modern Approach; Human Compatible (2019)
Across camps Build systems that stay uncertain about human objectives; a design constraint that cuts across every camp.