11

I remember a discussion where people were so sure in their ability to detect AI, when SE decided to not allow dealing with potential ChatGPT answers.

What I see in reality, occassionally, is that most people have 0 clue on how to detect AI, apart from the most obvious cases that are 90-100% copy pastes from ChatGPT. Here are some recent cases of nonsensical accusations:

  • my longer answer about captchas (accusation got reported and deleted, but downvote and delete vote stayed as of now).

The only suspicious things here are paragraph splits and decent enough English - but it's definitely not even perfect. Otherwise, a lot of phrases and wording are what AI would never use - "1st", "(= ...)", "have/want", etc. There's also some doubt/uncertainty which AI very rarely shows.

Any popular AI detectors gave this 0% rating, so no clue where the idea came from. These tools aren't perfect either ofc, but if they show no relation to AI at all, I don't know what's the point of trying to accuse someone. In that case AI usage is hidden so well that the idea is copied but the post is fully rewritten, which can't ever be proven. Even if contents are incorrect, as the poster could just be wrong himself.

Similarly, it's longer, it has paragraphs, good English and... that's it. It has even more syntaxes and wording that AIs wouldn't go for - unclosed (, "file/stream", quoted frames, "(1)" type of refences in the middle of text, etc. The ratings are yet again 0. No idea what encouraged someone to instantly write a comment about ChatGPT.

ChatGPT accusation

The fact that comments can contain basically anything and are only up for mods to remove if you report them doesn't help here. IMO bad comments should be more punishable, but that's another topic.

I'll just end this with a note that you don't just go to a store and accuse people of selling fakes with 0 analysis of the products. That's complete lack of manners and common sense and could actually lead to your arrest and a court case against you for defamation or anything similar. You also can't pick some random books in local library and damage them because you think they're copied. Internet is more free in terms of speech, but shouldn't be more free in terms of vandalism - yes, I think that attempting to discourage good quality answers in a mindless way is exactly that. Think about it.

2
  • Comments have been moved to chat; please do not continue the discussion here. Before posting a comment below this one, please review the purposes of comments. Comments that do not request clarification or suggest improvements usually belong as an answer, or in Super User Chat. Comments continuing discussion may be removed.
    – DavidPostill Mod
    Dec 24, 2023 at 9:59
  • 3
    "did you use ChatGPT?" is not an accusation. it's a neutral question. yes, some people are not as good at detecting AI-generated content. so you answered your own question. Dec 25, 2023 at 9:40

2 Answers 2

18

I get the slightest feeling of something that's turned from trying to keep the site 'neat' into essentially witchcraft trials.

As an end user - if you suspect someone of posting AI answers, there's literally no point in accusing them, or getting into an argument. If you're right, it's someone who clearly cares so little they're copy-pasting machine generated garbage. If you're wrong, you've gotten into an entirely unnecessary argument. It's probably better to focus on factual blunders in the question and flag for attention if you are reasonably sure it's an AI poster.

As to the why? Well, I'd say people are frustrated with AI - as a community-generated knowledge base that's done perfectly fine without AI, and TPTB don't understand 'why' people leave the place. People believe AI is the font of all knowledge when it's basically ... not, and they take its babbling for truth.

Unfortunately, we don't really have the support to deal with the underlying frustration causing these folks to rage - especially considering the 'recent' drama over SE trying to regulate moderation of chatGPT and such posts.

I'd say though, if you're an active user of the site, put down the pitchforks and torches and let us know so we can deal with it. Accusing people of using genAI and being wrong isn't nice. We have better tools - leaving it to the mods rather than arguing over it in comments is a better idea.

3
  • As for getting frustrated with AI - that's understandable. One of many many reasons you could get frustated with on this network. I'm not too sure if spamming/overusing "Is this ChatGPT?" or analogically "Is this answer wrong?", "Is this question bad?", etc. helps in any shape or form, though. I fully agree that reporting should be the preferred way to go if you're not 100% sure about something, so marking this as an answer for now, I don't know how much it's official SE stance on this matter or just personal opinion of a specific mod, though.
    – Destroy666
    Dec 24, 2023 at 12:55
  • The official stance is uh... rather confused I'd say. That said, my personal opinion carries some weight on this site, and if the official SE stance on this matter varies from what I feel is best for this site, well I'd have to hand in my diamond.
    – Journeyman Geek Mod
    Dec 24, 2023 at 13:14
  • 1
    "I'd say people are frustrated with AI" Don't include me in this, I'm fine with AI. Dec 27, 2023 at 17:45
1

The premise of the question in the title makes a very broad assumption regarding intent:

“Why do people attempt to discourage long well-written answers by instant AI accusations?”

Why do you think someone leaving one simple comment that reads, “Is it ChatGPT?” is doing so with the intent to discourage answers; well-written or otherwise?

What impact does one measly, disposable comment have on the large answer? If the answer is valid it gains upvotes and comments are meaningless; this happens regardless of any comments.

“IMO bad comments should be more punishable, but that's another topic.”

Oh, please give me a break. Nobody is leaving comments that suddenly destroy answers. Someone asking “Is it ChatGPT?” like I sometime do, are not out to destroy answers.

I have posted TONS of content here that is well written and thought out but does not read like LLM created content.

The biggest issue with your valid examples shown is that normal humans don’t post intent that reads “In conclusion…” and such.

Some people write essays that are not badly written but are just linguistically ridiculous. Cut down on the purple prose.

“None of this makes any sense to me and as can be seen by my post it leads to less knowledgable people spamming "Is it ChatGPT?" nonsense under many longer answers, which SE exactly wanted to avoid.”

What exactly are you complaining about? One or two lousy comments that ask “Is this ChatGPT?”

Like I said, I often leave such comments but never flag the answer or question unless it is truly awful. The thing that gets me are overwritten answers with headings and filled with purple prose. That is not good writing unless you are in Middle School.

That’s it.

Quit complaining and whining. One lousy comment should not drag down a valid question or answer.


PS: And to address this overly dramatic closing or your question

“Internet is more free in terms of speech, but shouldn't be more free in terms of vandalism - yes, I think that attempting to discourage good quality answers in a mindless way is exactly that. Think about it.”

If you are seriously that thin skinned that one lousy comment like “Is this ChatGPT?” is seen as “vandalism” then you should not post anything anywhere.

1
  • 1
    Upvoted for connecting "purple prose" to deep language models babbling. Which I feel in a Q/A forum like stack* should be avoided. I honestly also do not like answers coming from AI - they halluzinate randomly. So someone citing AI is someone I cannot ask for further clarifications or corrections, because he/she obviously is in no knowledgable position, too. Jan 4 at 10:58

You must log in to answer this question.

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