[VeryFinePrint]

As a reddit mod of a tiny community I've very recently started to suggest users use ChatGPT to re-word their comments and submissions when their phrasing is inflammatory. Any point worth making is worth making well. It is still early stages, so I'm not sure how effective it will be.

I've noticed that ChatGPT can do a decent job rewording an angry message to be less toxic while drawing out the salient points. The output it produces a little stiff or stuffy, but I find that preferable to "authentic" toxic messages.

As a mod I find that removals have less sting when you provide the user with an alternative path forward to get what they want. When a removal isn't just a "wall of no", it feels less like a personal rejection. ChatGPT offers a path around that "wall of no" for users by a) showing them what the right way to make their comment is and b) gives them a path/tool to getting there nearly every time.

[Teever]

It will be interesting to see when spiteful users start creating LLM powered bots that are designed to waste reddit moderator's time in such a way that the moderator's reaction doesn't rise to the level of a ban, but enough that it consumes a disproportionate amount of their time.

It'll be a fabulous display of malicious compliance where they're doing something somewhat similar to your suggestion of using chatgpt to rewrite their messages, but with an entirely different outcome than you intended.

[jmole]

It'd be hard to read this kind of a response as anything but passive aggressive, unless you are actually sending their comment to ChatGPT on their behalf.

"Please use chatGPT to fix your comment" is just too snide to be useful, like "RTFM" or "just google it".

[VeryFinePrint]

I send their comment through ChatGPT, then paste is as an example in the response. "Please use ChatGPT, this is an example of what your comment would look like..."

[gosub100]

This is interesting. But what I dislike about it is breeding the fallacy that GPT is the almighty authority and were just mere mortals.