In a guest post on Terence Tao’s blog, mathematician Frank Calegari argues that research evaluation and hiring incentives must change as AI produces proofs more quickly. He proposes assessing every paper as potentially AI-assisted, using direct interviews to examine understanding and research vision, and combining broader graduate education with independent technical training. Earlier reactions from researchers also stressed that checking and understanding large releases of claimed proofs takes time. The essay calls for less emphasis on publication counts; it does not establish that AI has completed mathematics.