Free Speech, Hate Speech, and Higher Education: A Theory.

From left to right: Claudine Gay, former president of Harvard University, Elizabeth Magill, former president of the University of Pennsylvania, Pamela Nadell, a professor at American University, and Sally Kornbluth, president of the Massachusetts Institute of Technology, at a fateful hearing of the House Committee on Education and the Workforce on 5 December 2023 in … [continue reading]

OpenAI, The Superalignment Problem, and Human Values.

A simple analogy for superalignment: In traditional machine learning (ML), humans supervise AI systems weaker than themselves (left). To align superintelligence, humans will instead need to supervise AI systems smarter than them (center). We cannot directly study this problem today, but we can study a simple analogy: can small models supervise larger models (right)? (Open … [continue reading]