A University of Chicago mathematician says AI-generated proofs could make academia rely less on paper files and more on personal interviews, communication skills, and a clear research vision when hiring mathematicians.