I recently had a disagreement with the lemmy.ml admins, specifically @AgreeableLandscape@lemmy.ml and @cypherpunks@lemmy.ml, regarding an LLM post. One mod claimed that it violated rule 4 (spamming), while the other insisted that I must attribute the content to the LLM. Typically, I use LLMs for most of my posts to correct grammar, spelling, and other aspects, or simply to generate the content given the basic thought. I am seeking an instance where administrators are accepting of LLM-generated posts and do not require any attribution, not even in a general manner like "LLM-generated post."

The admins did not remove the comment; a community mod did. Mods can impose further restrictions on their communities on top of instance-wide rules (within reason, of course), including banning LLMs. Lemmy.ml, at least, does not have a blanket ban on LLMs, but generally, it's expected that 1) you should not post LLMs excessively—mainly, we want to host discussions by humans, and 2) you should disclose that it's from an LLM and specify which one it's from. It is also preferable to add your own comments or analysis to the LLM-generated content. If it's a mix of LLM and your own writing, please mention it at the start of the comment. However, if the community directly disallows LLMs, then you shouldn't post them there at all.

— @AggreableLandscape@lemmy.ml

Thanks for clarifying. I'm deleting your generated comment per rule 4 (spamming), as well as two other generated comments you posted elsewhere. If another admin wants to undelete any of these, I would be surprised.

Please do not post LLM-generated comments without clearly labeling them as such. In my opinion, this is common sense and doesn't need its own rule. Rule 4 is sufficient.

— @cypherpunks

this has been gnerated by an llm and I woudl prefer no t to mentioan it in every pst I wrtie.

top 4 comments

sorted by: hot top controversial new old

[–] TheOneCurly@lemmy.theonecurly.page 14 points 2 years ago

I obviously can't speak for instance admins or mods but it really annoys me as a user when someone posts a giant wall of text that was clearly generated by an LLM. When I want information on a community like this I want real personal experience and knowledge. Nothing irks me more than someone starting a long comment with "I don't know the answer but I asked chatGPT and here's what it said". To me that says they have no knowledge that would let them edit or validate GPT's answer, so its useless to me.

[–] terribleplan@lemmy.nrd.li 7 points 2 years ago

I can't speak for this instance, but as an asshole with an internet connection (and my own small lemmy instance), I feel compelled to share my unsolicited opinion.

If the post is mostly or entirely your original thought and you only use an LLM to change editing/phrasing/grammar or to decrease the detail in your thoughts I don't thing any instance would have a problem with that. I say this because I doubt any user/mod/admin would be able to tell. You don't end your post with "edited by Grammarly" or "run through MS Office Spelling and Grammar checker". The thoughts are your own and the knowledge is your own (or something you researched while writing the post).

If, however, you tell the LLM "Write my a response refuting this post: [post text]" it is not you doing the thinking or having an opinion, it is the LLM. You pasting a response you did not write adds very little to the conversation. If anyone else wanted to know what ChatGPT/LLaMA/Bard/whatever "thought" about the post, they would ask that tool for a response.

If someone else wrote an article that refutes something someone posted I would comment and say "Sarah Whatshername wrote a great response to this where she mentioned that [... some info/quotes/whatever ...], I think it's worth reading before you buy into this too much" or something. In that sort of response I would be clearly crediting the creator of the answer I am adopting the opinion of as well as (if possible) linking to the source that I am referencing while writing it.

Without being able to see the removed comment, it is hard to say how you are using the tool, but based on the community reaction it seems like it leaned toward the LLM doing the thinking. This is basically what the admin said with the whole "preferably add to what it says with your own comments or analysis". If you yourself haven't used NixOS or read something actually written by someone who has used NixOS, then what do you have to add to the conversation by giving opinions or speaking authoritatively on it, especially if they aren't really your opinions but something an LLM has come up with based on however you prompted it?

I apologize if I come off as overly aggressive or negative, but in the thread that followed you were very combative after it has been pretty clearly laid out what is expected in that particular community and on that particular instance, so I am granting you less grace than maybe I should. You would be welcome on my instance or any of the communities I moderate, so long as all the actual thinking is done by you rather than by an LLM. If you were on my instance and people started reporting your posts (especially if admins started threatening to block my instance) I would need to review your comments and we would need to have a conversation about how you are using those tools, and why you are engaging with communities in an unwelcome way. If the outcome of such a conversation was you refusing to follow the rules of communities on other instances and refusing to adjust how you use your tools you would get the boot.

[–] TiredSpider@slrpnk.net 1 points 2 years ago (1 children)

maybe I'm ignorant but I don't know why you need an llm to correct grammar and spelling? plenty of programs already do that. I agree it seems spammy and I'd prefer not to see it.

[–] TiredSpider@slrpnk.net 1 points 2 years ago

UNLESS you are using it as an accessibility tool to cope with disorganized thoughts or other symptoms of mental disability that make writing difficult and you check the text after for misinfo. I could see the use in that.