I posted this first in What’s Up Today before @neo kindly pointed me here. Thanks neo.
Posting of ChatGPT and similar in block copy-paste (sometimes without even acknowledging that it’s non-human content) is apparently being normalised (here, on Discord), and I suggest we either have a rule disallowing it, or at least a policy on it. Other forums (I noticed stackexchange I think the other day) have explicitly banned it.
I think it adds nothing to the discussion, and is spammy. If a human asked ChatGPT a few questions, and then put the work in to think about the answer, remove loads of the irrelevant bits, and then summarize their thoughts, it’d feel far more normal than this block-pasting.
The etiquette in other forums (I’m thinking of Hacker News, for example) seems to have naturally swung towards: obviously don’t block-paste a wall of unedited content. Here that etiquette doesn’t seem to be naturally taking hold, so perhaps a statement on the matter, or a rule, is necessary. Cheers.
EDIT: should probably tag the @moderators, for their attention.
Enforcement will involve asking when it seems likely, and we do get pretty obvious examples where the content isn’t introduced, and isn’t connected to the conversation.
If someone tries to pass off generated content as their own words of course it’s harder to spot, but if the content is actually good enough to pass as a normal reply, then it’s much less of an issue anyway.
The problem, for me at least, is when we see a big wall of text that isn’t part of a discussion. I skip over them because I understand what the output looks like well enough to spot it. But it is disruptive and wastes every reader’s time, usually without contributing value.
I don’t understand why people do that without explaining that it was generated, and without explaining why they’ve included it, but they do.
I’d prefer it was excluded, but it should at least be quoted, attributed, and ideally the reason for including it explained.
We could possibly formulate a suitable forum guideline that allows us to deal with anyone or anypost that doesn’t attribute large quotes or LLM output to being that.
Maybe @JPL could help with that since he is good at englishing