Safety2h ago
Virtue Ethics Proposed For AI Alignment
A new essay in The Gradient argues that rational agents should lack fixed goals. Instead, it proposes aligning AI to practices—networks of action-dispositions and evaluation criteria. This shifts the alignment focus from goal-optimization to virtue-ethical agency. Practitioners should consider this framework to avoid the rigid failure modes of traditional objective functions.