Posts

AI models inherently alter "human values." So, alignment-based AI safety approaches must better account for value drift 2025-01-13T19:22:41.195Z

Comments