-12

Regarding the SE sites allowing the use of generative AI:

https://physics.stackexchange.com/help/ai-policy, https://webapps.stackexchange.com/help/ai-policy, https://serverfault.com/help/ai-policy, etc. state:

You may want to share details on the prompt you used to produce the output too, but there is no need to completely copy and paste the prompt and output. Just as when referencing any other content you didn’t author, you should generally avoid copying the complete text, and should instead use the words and ideas from this content to support your own.

and give the example:

I asked [Generative AI service] about [partial prompt]. Its output was:

[Generative AI response]

[other sources, quotes, explanations, etc. necessary to complete the post]

Feature request: In the AI policy, don't discourage sharing the entire prompt. On the contrary, it should be encouraged.

Motivation: having the entire prompt makes it easier to improve on it, ask for follow-up prompts, and learn how to optimally prompt.

Improve this question
12
  • 7
    While the base text was initially copied to many sites, this seems like an issue that should be handled by each individual site, as policies will differ. I'd suggest you bring this issue up on the sites where you'd like to see this happen and get buy-in on those individual meta sites. Once you have that buy-in on that site, a moderator on that site can edit the text for that site.
    – Makyen
    Commented Jun 2 at 20:54
  • @Makyen I suggest changing it for all the SE sites allowing the use of generative AI.
    – Franck Dernoncourt
    Commented Jun 2 at 20:56
  • 9
    I understand that. However, past the point that it was pushed out to all the sites, SE has specifically stated that the moderators of the individual sites are the ones who will be able to modify that page. You're asking SE to take back control of that page and push out an update which each individual site might not want, and which is not an overall AI-policy issue, but is an issue which is likely directly a per-site flavor issue. What I'm saying is that this isn't an all-SE issue. It is a per-site issue.
    – Makyen
    Commented Jun 2 at 21:13
  • One issue I've encountered is that links like this seem to be broken (at least, they don't work at my end). And giving the entire "prompt" (actually a prolonged conversation) is impractical.
    – Rebecca J. Stones
    Commented Jun 2 at 23:28
  • @RebeccaJ.Stones ChatGPT users may clear their chat logs (I did it once accidentally). Best to encourage sharing the entire prompt in SE posts. Your link works though i.sstatic.net/VlcwXIth.png Mosts prompts are <1k chars.
    – Franck Dernoncourt
    Commented Jun 2 at 23:29
  • 1
    Oh, thanks for letting me know! If the links work, then I think it's reasonable to link to the entire conversation where relevant, basically for the same reason as you might link to a log file.
    – Rebecca J. Stones
    Commented Jun 2 at 23:42
  • @RebeccaJ.Stones in most cases, one only needs 1 rather small prompt. Better copy it on SE as links may get 404.
    – Franck Dernoncourt
    Commented Jun 2 at 23:47
  • 2
    My interactions with ChatGPT are conversational, like this. Private instructions and earlier interactions (and typos, and version) influence how ChatGPT replies. E.g., a "prompt" like "And how does 传递 fit into this?" cannot make sense out of context, and ChatGPT's response could not be reproduced without further information. (Besides, a link is substantially less obstructive than a full prompt.)
    – Rebecca J. Stones
    Commented Jun 3 at 0:10
  • 4
    Wouldn't such a thing put a disproportionate emphasis on the AI-generated content, and allow much more room for completely irrelevant comments? This would mostly be interesting for those wanting to hone their promoting skills, which SE is not the platform for.
    – Joachim
    Commented Jun 3 at 5:30
  • 2
    @Joachim Explaining how one was able to find an answer makes the answer more useful. SE is a platform for learning.
    – Franck Dernoncourt
    Commented Jun 3 at 5:34
  • 1
    Certainly, but not for learning how to find answers. And I'm sure it would be fine to ask for a prompt, or for some users to share it anyway (within reason), but it seems logical to dissuade users from posting unnecessary long or detailed commentary on their path to enlightenment.
    – Joachim
    Commented Jun 3 at 5:41
  • @Joachim Most prompts are short. I think it's better to help OP learn how to find answers.
    – Franck Dernoncourt
    Commented Jun 3 at 5:44

1 Answer 1

Reset to default
4

I suspect most readers would prefer answerers to cut to the chase rather than narrate their research process. The general ethos of S.E. sites is to pare away anything extraneous from answers, so the current guidance is fine as a default; all the same, if any community feel differently they can edit the help page to encourage what they like, as @Makyen points out

Anyone with a particular interest in prompt engineering might care to visit https://genai.stackexchange.com

Improve this answer
0

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .