Dr. Yusuf Okafor, 41, AI alignment researcher who published three foundational papers on mesa-optimization before concluding that corrigible AGI is theoretically impossible given current training paradigms. He still writes code, for startups now, because the money is good and the problems are finite. He is cheerful about civilizational risk in a way that makes his colleagues uncomfortable. He believes intellectual honesty requires facing conclusions even when the conclusions are bad.