Can a uni reliably say someone has used ai to create their work?

[Unissss]

i personally don’t see how tbh

I think it's a bit more complicated than accusing people of using AI because of em dashes or using the word 'delve' - Word itself sometimes changes hyphens to em dashes. But I don't doubt there's usually a clunk somewhere. I'm worried for people not using AI but being flagged for it because of their writing style. The implications for ultimately dumbing down on that basis are worrying.

This would absolutely not fly. Special adjustment requests for e.g. anxiety have skyrocketed in the past 5 years and there would be huge pushback.

My daughter was accused of this along with others (she would never!!) last year in year 7 because her writing was "too good" for her age. She was then put on the spot and made to do a half hour creative writing session. The teacher then commented in writing something alone the lines of, oh you really ARE talented etc etc etc it went on quite a bit. Some of the others admitted it or were found 'guilty' even though they denied it.

if she had had an off day with her writing or whatever, I truly believe they would have followed through with their accusation.

So no, I don't think whatever little method someone might have are always reliable.

— is no more convoluted than any other shift key operation (such capitalisation or a currency sign) on a Mac and most modern keyboards allow macros to be set for symbols not on the default layout (which is what I do if using Windows).

Any cheating pupil with a brain using AI will know the keyboard strokes, at least on their own machine. They may genuinely not be able to reproduce it on a different machine if they are using macros.

Its true that modern checkers are much more reliable than a few years ago but its also very easy to confuse certain types of word structure with AI.

As @SerafinasGoose says upthread - the only way to be sure is to interrogate them on the content.

There is telling (easy in most cases), and then there is proving (impossible).

My DS is at uni and they are supposed to sign a sheet saying they’ve not used AI to write their work. I suppose if a student turned in an assessment with a very different writing style to normal, it’d be flagged. You can get thrown out for using it, so not worth the risk imho. You can use it for research up to a point though.

IANAL but also frequently have to write complex technical content in a structured format, for less technical audiences and with one eye to contractual obligations. I also use em dashes quite a bit - this is a reflection of my age rather than AI tools.

My minions had quite a lot of fun putting content I had written many years ago through an AI checker 😀

Checkers are better than they were but the only reliable test is actually talking to the author to test their understanding of the content and how it was developed.

To add. Not so much a viva, but on DS’s UG course they all had to ‘present’ their dissertation and answer questions from the lecturers on it. I suspect that’s a reasonable way of spotting anyone who had their work written by AI or otherwise.

To a degree, but all AI detectors throw up loads of false positives, especially if the writer is particularly accurate and polished in their style - AI believes humans are all inconsistent and make errors.

I would literally rather sit in a gym hall and write the whole thing my hand with ten drafts than do any kind of viva 😅

People talk about “knowing” students. But on large modules ( e.g. 500) we can’t. Marking is anonymous anyway and often carried out by others e.g postgraduate teaching assistants.

Grammar and spell checks can reformat work even if you don’t ask it to.

I’ve never used it for that but I had a wall of audio transcribed text that I asked AI to add sentences and paragraphs to. It took several goes to stop it re-writing the text as well in its own classic AI speak.

I'd hand deliver it, if someone else wants to scan it and convert it, they can do - I wish AI the best of luck deciphering my handwriting.

The aim of handwriting it would be so that I did have proof of the essay's evolution, rather than an electronic document that didn't show revisions and corrections. I suppose someone could prompt AI to produce a realistic 'notes, drafts and corrected drafts' documents package but we're getting into the realms of it taking longer to write prompts that work than it would be to write the essay oneself.

On the contrary the shift back to exams has already happened.

Typing and special adjustments is different - although I think may be applied across the board eventually.

I don't trust them, I ran a report through that I had written for work (all my own work!) and it estimated 30-35% AI

Because there's only one (Pangram) that works well. You can't just type 'AI text checker' into google and just believe that they're all flawless. And you can't see one working badly (or well) and assume all are the same. Literacy around what AI can and can't do, and which platforms are reliable, is a big issue.

Why would you use AI for that? A Word document with spellcheck switched on would highlight spelling and grammar errors.

Yep same, all our exams are now in person. Obvs with adjustments where necessary, but invigilated.

I know that lecturers can have very strong suspicions that AI was used. If a student admitted it when questioned or there was concrete evidence like a prompt left in the assignment, then their is clear evidence. Potentially multiple fabricated references could also be used as evidence. However, beyond these cases I would have thought the application of academic sanctions could be challenged by an internal complaint and ultimately in court.

An academic may suggest that a discrepancy between oràl explanations and the written assignment points to AI use (and are probably right). But would this academic judgement be accepted if challenged? The student could say they were nervous and/or impacted by additional needs when questioned. What level of discrepancy is sufficient? Also unless an oral defence, when there is suspected AI, is part of a formal policy students could argue it is unfair for them to have to engage in this additional assessment process. What is the threshold to trigges it? How is applied in a way that is consisent across the cohort?
I am curious whether students have challenged sanctions in situations where there is no admission of guilt or other clear cut evidence.

Same!! I am much more articulate when writing than when speaking

Discrepancy between writing style and oral style would be an utterly ignorant hook to hang sanctions on.

My writing style in essays and work reports is totally different from my speaking style. It's just the way my brain works. (By contrast my daughter, who is dyslexic, is the polar opposite and far more articulate when speaking. She'd ace all her exams if they were oral!)

Unfortunately I made a very stupid mistake and where I got it to correct my spelling I left the prompt in it unfortunately. The prompt does say would you like to make it more academic style so that would surely make it believable that I only used it for spelling?

Are you sure you’re at uni? Because you came on here asking if they have any way to tell, then you said that you just don’t see how they could tell, then you said you’d been accused of it so this thread was you trying to find out how they could possibly know, then you admit that you did use AI and now you’ve admitted that you left the ChatGPT prompt on your essay when you submitted it. Do you see how these questions of yours are not logical?

You’ve been saying that you just don’t believe there is any way they can tell… when you left the same prompt in it! That’s just not a logical thought process.

They have proof you did it. If you are telling the truth that all you did was check spelling then take your laptop in, and show them your save history to show the progress of you writing the essay, show them the entire conversation with ChatGPT so they can see all you did was ask it to check your spelling. If you are telling the truth.

If you used AI to write or edit your essay, and you’re lying to us then this is an entirely pointless thread. It pretty much already was because you left the prompt in. Be prepared to fail this class. And start being honest.

I hadn’t read through it as I submitted in a rush in the early hours of the morning. I’ve now just had the time to read through and saw it. Ive already stated I used it for spelling didn’t say I didn’t….

I am honest I’ve already told them what I used it for before I even saw the prompt so. No need to make me outro be a dishonest person we all mistakes and sadly this is one of mine. If I was cheater type I’m sure I would have done before the last assignment of my degree.