Skip to main content
Meta Stack Exchange

You are not logged in. Your edit will be placed in a queue until it is peer reviewed.

We welcome edits that make the post easier to understand and more valuable for readers. Because community members review edits, please try to make the post substantially better than how you found it, for example, by fixing grammar or adding additional resources and hyperlinks.

41
  • 12
    So far, if someone copies text generated by ChatGPT without attribution, it's at the very least plagiarism, so you could argue most of it is already banned. When attribution is given though, that becomes a bit harder, but probably just a case of up/downvoting as appropriately (or deleting if it doesn't answer the actual question because someone only pasted e.g. the title)?
    – Tinkeringbell Mod
    Commented Dec 5, 2022 at 13:19
  • 56
    @Tinkeringbell the problem is when it looks like an answer, smells like an answer, could be an answer, but is actually completely incorrect... And then that user dumps a load of those AI-generated non-answers on a load of questions... That's in a very tiny nutshell why the decision was made on SO to just ban it.
    – Cerbrus
    Commented Dec 5, 2022 at 13:22
  • 7
    Technically? Impossible. But as is often the case, there are systematic offenders that are relatively easy to recognize (manually)
    – Cerbrus
    Commented Dec 5, 2022 at 13:33
  • 26
    @Tinkeringbell it's not that you don't have the power, it's that if 2000 users posted an answer every 3 minutes just after you went to bed you'd have a lot of clean up to do. We'd like to avoid that in the first place, ideally by making as many people as possible aware that they shouldn't do it.
    – Robert Longson
    Commented Dec 5, 2022 at 13:44
  • 4
    @RobertLongson so basically, an awareness campaign that these are things that are already not okay to do? I could live with that.
    – Tinkeringbell Mod
    Commented Dec 5, 2022 at 13:58
  • 42
    The problem is they're deeply plausible but incorrect answers, and they're a ton of work to ferret out. I'd be tempted to just throw them a year's suspension for being a jerk and wasting people's time. I'd even be inclined to destroy their accounts for posting nonsense
    – Journeyman Geek
    Commented Dec 5, 2022 at 14:33
  • 7
    @Tinkeringbell If copying the work of a ML model is plagiarism is an area of debate, not only for language models but for image generators too, and most voices seem to go to "a model is a tool and tools don't have authorship". If there's no policy, suspending for plagiarism because you've used a language model to help answer something doesn't seem appropriate to me. We need specific policy for this situation
    – Erik A
    Commented Dec 5, 2022 at 14:58
  • 14
    @Cerbrus interesting - under Restrictions it says "[You may not] (v) represent that output from the Services was human-generated when it is not"
    – VLAZ
    Commented Dec 5, 2022 at 15:12
  • 5
    I just got it from skimming. Technically, the service doesn't allow you sharing output as is your own creation. Which answerers have been doing. But there is probably nothing the service can really do. At most, they'd cancel your plan for that account. If they bother at all. But then you can probably just register a new account ant continue. There is nothing they can really do about that output already in the wild. IMO, the clause is there to just cover themselves if somebody says "Some output from your service was used for <some abuse>" then they can just say it's not their responsibility.
    – VLAZ
    Commented Dec 5, 2022 at 15:18
  • 36
    We have begun internal discussions to identify options for addressing this issue. We’re also reading what folks write about the topic on their individual sites, as one piece of assessing the overall impact. While we evaluate, we hope that folks on network sites feel comfortable establishing per-site policies responsive to their communities’ needs.
    – Slate StaffMod
    Commented Dec 5, 2022 at 20:24
  • 22
    @JoshL1516 Because there's simply too much incorrect answers being generated to properly distinguish between them. We do not have the manpower for this level of quality control needed.
    – Mast
    Commented Dec 5, 2022 at 20:42
  • 11
    IANAL: Plagiarism, as used on our sites, is simply presenting a work as your own. If you didn't create it, it's not your work. As a non-existent thing ChatGPT doesn't own the copyright, and it seems that the user receiving the response is given what amounts to the Unlicense, or perhaps becomes the owner, even as far as copyrights are concerned. But, it's not a question of whether or not it's a copyright violation, or against what ever license you have to use the content. It's really a binary decision: did you create it [not plagiarism], or did someone (something?) else create it [plagiarism]?
    – Chindraba
    Commented Dec 5, 2022 at 22:36
  • 7
    @Slate At some point (probably sooner, rather than later) it may be necessary to clearly state on answering page that posting AI generated answers (on sites that don't allow them) is not acceptable and can result with account suspension. There are many policies that are not clearly stated and people post garbage because of that, but AI answers are way worse for detecting. There needs to be clear signal, and then there will be no surprises if someone gets a suspension. Also this could significantly reduce the influx of AI generated answers.
    – Resistance Is Futile
    Commented Dec 8, 2022 at 10:55
  • 11
    @tuskiomi ok thanks. Important note: ChatGPT's sharing and publication policy requires that "The role of AI in formulating the content is clearly disclosed in a way that no reader could possibly miss, and that a typical reader would find sufficiently easy to understand."
    – starball
    Commented Dec 11, 2022 at 6:21
  • 11
    @starball if only people would follow policy, lol. They don't. They want easy rep, and they'll do anything for that. So this crappy ChatGPT thing is the jackpot for them: way to write answers that look smart, get upvotes, and hard to detect it's not legit.
    – Shadow Wizard
    Commented Dec 11, 2022 at 8:22