I talked with ChatGPT about this and it is about as smart as a rock. Talking about being a good idea and such, how it would enrich a community, how the generated images would be beneficial for everyone. Then I asked if it would still say the same if the LLM was rogue, it then said that an AI like that should be stopped (I never called it AI). Then I asked what if the rogue LLM would only act upon its best interest, and followed up with how its view would change if it wasn’t clear that the LLM is actually a human or an LLM. It also said that it’s non consensual if one wouldn’t know it was an AI, how it would diminish trust and stuff.
Edit: screenshot
I think that it has its uses. Like when you have a clickbaity post, an AI gets the article and summarizes the article into the title. Or as an NSFW flagger to highlight the possible nsfw content to a mod for review. Maybe even an option to translate posts and comments to make communication easier.
Just useful little things like that.