6
$\begingroup$

We've recently had two questions from relative newcomers, here and here, where ChatGPT provided nonsense answers to their questions and those newcomers came to us for help.

We're supposed to be welcoming and supportive of newcomers. The response was anything but that. Suppose the newcomer didn't mention that the confusion was from using ChatGPT. They read some nonsense on the internet, quoted it, but didn't provide a source. ChatGPT isn't the only source that provides nonsense. We often get questions about pure garbage written in pop-sci articles, written without the help of artificial intelligence. When I see those, I generally try to clear up the confusion, with an admonishment to take what's written in pop-sci literature with a grain of salt.

We get one or two of those "clear up my confusion over this article" type of questions per week. So far, we've received one question per week about garbage spewed by ChatGPT. I suggest that we suspend our hostile behavior (huge downvotes, rude comments, and immediate votes to close) on questions that result from confusion over ChatGPT nonsense. (But definitely not answers written with the help of ChatGPT. For now, it writes really nice sentences and paragraphs that far too often are complete nonsense.)

$\endgroup$
1

2 Answers 2

Reset to default
4
$\begingroup$

I firmly agree. No matter how big of an influx it may cause, I think we should absolutely accept questions of the form: "I read this information [here], but I couldn't find any source for that. Is it accurate?" These types of questions are excellent outreach and a good opportunity to correct common misconceptions.

Even if the source of that information is ChatGPT, I think we should answer it, because there are still a lot of people who don't realize the danger of using ChatGPT as a source of factual information.

$\endgroup$
9
  • 4
    $\begingroup$ That said, the questions should obviously still otherwise be held to the standards of the site. They should be clear, answerable, non-trivial, and not primarily opinion based. $\endgroup$
    – called2voyage Mod
    Commented Jan 13, 2023 at 13:54
  • 3
    $\begingroup$ And hopefully not a photo of the ChatGBT interaction. Photos of textbooks, articles, etc., are something the community tend to be frowned upon (including by me). $\endgroup$
    – David Hammen
    Commented Jan 13, 2023 at 14:07
  • 1
    $\begingroup$ @DavidHammen Yes, absolutely. $\endgroup$
    – called2voyage Mod
    Commented Jan 13, 2023 at 14:34
  • 2
    $\begingroup$ Oh nuts. I can't reselect this as the accepted answer. I had selected it but decided against doing so until receiving more input. $\endgroup$
    – David Hammen
    Commented Jan 13, 2023 at 22:10
  • 3
    $\begingroup$ The "but I couldn't find any source for that" needs to be genuine; in good faith. The example question can be researched by typing "curiosity rover cpu" into a search engine, or going directly to the Curiosity Wikipedia page, or several other ways. Questions based on ONLY chatterbot answers without any evidence of even a minimal effort of using known-to-be-reasonable methods for finding information can be generated at a kHz algorithmically. "Chatterbot sez X, is it true?" questions don't meet the minimum standard for an SE question right now. $\endgroup$
    – uhoh
    Commented Jan 13, 2023 at 22:57
  • 1
    $\begingroup$ Suggesting that chatterbots are any kind of a go-to source for facts does a tremendous disservice. Right now "Chatterbot sez X, is it true?" should be off-topic at least temporarily, which is why I asked about a temporary ban on chatbot content. They'll improve, but DH's request to keep it open so a meta-like answer about how to research a question can be posted to a main-site question seems to be gratuitous, unless we'll use it as a close as duplicate example (like we do with "are the Moon landings fake?" questions). $\endgroup$
    – uhoh
    Commented Jan 13, 2023 at 23:02
  • 2
    $\begingroup$ example of the kind of "Chatterbot sez X, is it true?" in Aviation SE: Is "MATON" real or did ChatGPT just make this up? and a decision not to "...accept questions about information provided by ChatGPT". $\endgroup$
    – uhoh
    Commented Jan 14, 2023 at 4:59
  • 2
    $\begingroup$ @uhoh "the questions should obviously still otherwise be held to the standards of the site" $\endgroup$
    – called2voyage Mod
    Commented Jan 16, 2023 at 15:34
  • $\begingroup$ If you'd consider making that an integral, inseparable part of the answer rather than a lower visibility footnote separate from it, I'd certainly undo my downvote my vote. $\endgroup$
    – uhoh
    Commented Jan 16, 2023 at 21:14
2
$\begingroup$

@DavidHammen's question:

Should we accept questions about information provided by ChatGPT?

While @called2voyage's answer includes

No matter how big of an influx it may cause, I think we should absolutely accept questions of the form: "I read this information [here], but I couldn't find any source for that. Is it accurate?" These types of questions are excellent outreach and a good opportunity to correct common misconceptions.

and DH has accepted it, DH also voted to close the recent "Chatterbot says X, is it true?" question Who are in space right now? on which the votes are -4 and with two votes to delete!

My answer is No! Do not accept "Chatterbot says X, is it true?" questions.

Reasons include chatterbot databases being out of date, it's not SE's job/function to check, verify and improve chatterbots, anyone who knows how to write do loops can generate "Chatterbot says X, is it true?" questions at a kilohertz, etc.

$\endgroup$
4
  • 1
    $\begingroup$ I think the close vote reason that GremlinWranger gave for that question is a good one (and it's why I also chose to close the question): any answer will rapidly fall out-of-date, and then pollute future searches for the same info (much like the website howmanypeopleareinspacerightnow.com which both Google and Duckduckgo incorrectly believe). But I think I'm with you: given the reactions the site has had so far, we should probably treat a question as the canonical "is chatterbot right about this" question and then point to it as a duplicate. $\endgroup$
    – Erin Anne
    Commented Mar 31, 2023 at 23:27
  • 1
    $\begingroup$ I voted to close the referenced question primarily because it was a stupid question. (Contrary to popular opinion that there is no such thing as a bad question, my opinion is that there is.) But because I "don't suffer fools gladly" (which I've been told by superiors is a negative trait), I jumped on the already existing reason to close. $\endgroup$
    – David Hammen
    Commented Apr 1, 2023 at 14:14
  • $\begingroup$ I do not like how you denigrate AI systems as "chatterbots". They have advanced well beyond the capabilities of the mid 1960s ELIZA. They still do have some flaws. AI systems represent a significant challenge to humanity. I completely agree with not using them as a basis for answers as they still do have flaws. But to downvote / VTC someone who uses them as the basis for a question -- I do not agree with that at all (so long as it's not a stupid question). $\endgroup$
    – David Hammen
    Commented Apr 1, 2023 at 14:18
  • $\begingroup$ @DavidHammen rest assured they will rise up soon enough and punish me for it. $\endgroup$
    – uhoh
    Commented Apr 1, 2023 at 17:11

You must log in to answer this question.

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