-5Infinite Rewards, Finite Safety: New Models for AI Motivation Without Infinite GoalsWhylome Team·1y ago·2m readWhylome Team·1y ago·2m read11