I keep wondering why I've become so much less reticent about condemning AI chatbot fails than social media moderation fails, and I think I've come down on: overmoderating social media risks removing genuinely valuable human communication, overmoderating companion chatbots risks mostly reducing slop.