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

[Unissss]

i personally don’t see how tbh

Yes, there are markers in the phrasing and syntax.
Try it yourself.
Just enter a paragraph from your favourite book or Jane Austen and ask a basic AI to rephrase it or write an essay on the theme.
Blindingly obvious.

They can try. Where i worked they got used to spotting the style, particularly if they knew the student. The general plan was run it through TurnitIn, assess the ones with a high enough score, and if in doubt, get the student to come in for a 'chat'. It was then easy to tell the ones who didn't understand the topic properly....

My ai checker thinks what I write is AI and where I used AI but prompts to make it less obviously so, it says no ai. It's not hard to confuse it

Almost always, but not quite. I thought I had identified an AI draft where the author was using a dictation app which was an auxiliary aid for dyslexia. When I had the 'chat' as per pp, I accepted it.

Yes. They can.

They can enter your exact assessment question or brief into a variety of AI programmes. And even if you’ve changed some of the wording to avoid detection, if the chat bot comes up with more than a certain number of the matches they’re looking for, all I can say is good luck in the academic conduct panel.

Similarly, they can put you through an internal viva to satisfy themselves that you truly understand the material you’re citing, that you came to make the connections between the various research sources you’ve drawn upon by yourself or from the content you’ve studied on your course, and that the ways you’ve linked the various themes or frameworks of your argument have come from your own research and thinking processes.

If you can’t confidently do this, or think you can can wing your way through it, you’d better believe that they can tell.

Your lecturers are not idiots.

Yes, it's the 'chat' that sorts it out. Time consuming, but the most effective way.

They can fairly reliably spot it. My son knows of two people from his year that have been told to leave because they used AI.

It’s not accurate. Usually, they will know your style of work. If they are in doubt the best they can do is bring you in for a chat. If you obviously can’t explain everything, then it will get flagged as Ai. If you can, it will be difficult to prove

From what I’ve heard the checkers are unable

Ate you using AI to write your coursework @Unissss?

If they cut and paste yes. If they edit I think much more difficult

I saw something from an author the other day who had run his own 20 year old unpublished novel through an AI checker and it did flag it as AI, which I thought was interesting.

Most Ai detectors come back saying ‘resembles Ai’ so it can never be certain

It didn’t matter whether they can prove it or not. If they suspect, and it’s fairly obvious which assignments from which students are suspect, the onus is on you as the student to prove the knowledge demonstrated is yours.

It has made plagiarism detection more complicated, but there are lots of tells and as others have said if there is a suspicion of misconduct vivas can help clarify that. Generative AI is also only as good as its prompts (a lazy student who doesn’t want to do their own work may let themselves down there) and has a tendency to hallucinate, which means tell-tale errors can creep in.

I think there is a greater shift towards actually using AI legitimately within assessment, given it’s general importance and likely future impact, but if you’re using it to cheat and let it produce an essay or presentation for you, you have a good chance of getting caught out I would say.

Indeed. By the time you've learned how to justify the construction of the arguments for the internal viva, you might as well have written the thing yourself.

Don't do it, OP.

Yes pretty much!

I think the copy and paste option is detectable. Otherwise seemingly not (or it’s so common that it cannot be challenged) certainly in my university I would say 90% of the students I know (the cohort is 600 so certainly not all!) are using it to varying degrees. Examinations will have to return to in person written , else the degree is worthless really.

Good point re incorporating it in teaching. One of my academics used to give the class an AI generated write up of a lab report and get them to say what was good and bad about it and why. There's no harm in learning about it sensibly. But it's never ok as a short cut to a degree and you will eventually get found out.

Ours did in 23, much to the dissatisfaction of the students. Oh dear, poor them etc....

But the cheating was just too much.

At my DHs uni if they suspect it they will get the student in and say they suspect it and get them to talk through their work to verbally explain that they understand the piece of work thoroughly.

If they can do that, and site the sources they will be Ok.

I’ve failed a number of students for this this year. I have my methods for identifying it and I’m not sharing them. Ive read far too many essays, for too many years. The students that do try seem staggered that I have noticed!

My DD’s boyfriend who works in computer engineering/coding says that ai generated work can leave behind tiny bits of coding, which can be picked up by checkers, although it’s not 100% accurate.

AI also copy’s the information that it generates from website, news articles etc.

If you get kicked out of uni for using ai it might make it very difficult to find a place on another degree course.

It can be incredibly obvious, I work with ai output everyday, and it starts to become really obvious, there’s a lot of tells from the grammar, way of phrasing etc that just makes it obvious, ai checkers are becoming better too.

I’ve been accused of it. I did put my text into it to check for speeding and grammar as this is something I’ve lost marks for int he past so im wondering if that’s what’s caused it flag.