6
$\begingroup$

I was reading this question and saw that it started by saying

ChatGPT was providing a glimpse of the 8 possible gluon color states:

This horrified me as current AI models are definitely not a good source of information (even if in this case it was correct). So I decided to proposed an edit to the question to get rid of any mention of GPT. I changed it to

The 8 possible gluon color states are:

Now, my questions are:

  • Is citing AI as your source against the policy of this site?
  • Should I always edit questions that cite AI as their source, to disincentivize other users to ask more questions with similar sources? (Maybe I am incorrectly assuming that this can lead to bad quality questions)
$\endgroup$
4
  • 2
    $\begingroup$ I usually leave a comment telling the user that if they mention that they got this information from chat GPT, they will be heavily (and rightfully) downvoted. Usually they then rework the question on their own. $\endgroup$ Commented May 31, 2024 at 17:20
  • 2
    $\begingroup$ @AXensen Note that physics.stackexchange.com/help/gen-ai-policy permits GenAI content, but it must be properly referenced. $\endgroup$ Commented Jun 1, 2024 at 16:45
  • 1
    $\begingroup$ @PM2Ring Yes thank you you're right... to be more clear, if you get information from chat GPT (or any source), you need to cite it. I wasn't recommending that users remove the citation. Rather, when I come across "chat GPT told me this, is it right" questions, I recommend that the users delete the GPT generated content, come to their own understanding of the physics using conventional sources, then ask a question about what they haven't understood from those sources. $\endgroup$ Commented Jun 3, 2024 at 20:00
  • $\begingroup$ @AXensen Ah, ok. That's good advice. $\endgroup$ Commented Jun 3, 2024 at 20:14

1 Answer 1

Reset to default
9
$\begingroup$

The linked question seems like a good candidate for an exception to our broad preference to avoid chatbot/LLM-generated text in questions and answers, if only because the chatbot seems in this case to have generated correct information which is generally available from other sources, and the question itself is about the physics content of this correct information.

Note that the linked policy post reminds users that text from sources other than the author of the question should be attributed. Text from a sketchy source doesn't become less sketchy if you keep the source a secret; it's just harder to identify the problems with it.

If you want to guide users not to use chatbots as research tools in this way, you should be more direct about it. (Beware that some people believe that "direct" and "polite" are mutually exclusive, both as writers and as readers, and proceed with caution.) You can link to the policy post above discouraging computer-generated text, and you can remind the user that these "intelligent" tools are unreliable for questions on sophisticated or niche topics. Proposing alternative research avenues ("look here instead") is more helpful than just saying "don't use a chatbot."

$\endgroup$
8
  • $\begingroup$ Still, should I edit out any mention of chatbots, even after telling him that it would be better to avoid such sources? $\endgroup$ Commented May 28, 2024 at 19:54
  • 15
    $\begingroup$ You should not remove citations. Our policy is that sources should be cited. $\endgroup$ Commented May 28, 2024 at 20:50
  • 4
    $\begingroup$ We currently have a Help page physics.stackexchange.com/help/gen-ai-policy which states: "Generative artificial intelligence (a.k.a. GPT, LLM, generative AI, genAI) tools can be used to generate content for Physics Stack Exchange, but this content must be properly referenced as per our guidance." $\endgroup$ Commented May 31, 2024 at 21:36
  • $\begingroup$ @rob IMHO, the help page I linked is not consistent with the policy stated in Please don't use computer-generated text for questions or answers on Physics. Other sites have a GenAI policy Help page that clearly and strictly prohibits such content, eg math.stackexchange.com/help/gen-ai-policy Many members of Physics.SE assume that the policy here is along those lines, after the meta discussions we had in your linked question & physics.meta.stackexchange.com/q/14438/123208 $\endgroup$ Commented Sep 30, 2024 at 17:47
  • $\begingroup$ In light of those meta discussions, and discussions in The h Bar chatroom, I naturally assumed that Physics.SE would opt for the "No GenAI" help page option mentioned at meta.stackexchange.com/q/396626/334566 I was surprised (and dismayed) to discover that wasn't the case. I found out by accident, when I went to post a link to the Help page in a comment on a question that was derived from a ChatGPT interaction. $\endgroup$ Commented Sep 30, 2024 at 17:54
  • 1
    $\begingroup$ @PM2Ring I don't think we ever had a Meta discussion about the help center text. As I said to you in another context, I think the current help center page gives guidance which is broadly consistent with my answer here and with the other Meta discussions you've linked. But I can see how you would respond differently. It'd be appropriate for you to start a fresh Meta discussion specifically about the help-center text, so that the community can articulate its stance. $\endgroup$ Commented Sep 30, 2024 at 18:22
  • $\begingroup$ Note that we don't need to slavishly follow one script or the other; most pages in the help center can be edited by moderators. $\endgroup$ Commented Sep 30, 2024 at 18:25
  • $\begingroup$ @rob I've just posted physics.meta.stackexchange.com/q/14801/123208 $\endgroup$ Commented Oct 1, 2024 at 9:49

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.