Would you actually trust AI to moderate your forum?

Automated moderation is getting smarter, but handing full control to AI still feels like a gamble. On one hand, tools like LLMs and sentiment analyzers can filter blatant spam, hate speech, and toxic posts 24/7 without getting tired.

On the other hand, AI notorious for missing context, sarcasm, and nuanced banter - which are huge parts of natural community culture. Over-moderation can quickly kill user engagement, while under-moderation lets bad behavior slide.

Where do you draw the line? Are you comfortable letting AI auto-flag and auto-delete posts, or should it strictly remain a assistant for human moderators?