Use of AI to write assignments

[Origamiheaven]

This year university where I work has seen a huge increase in students using AI to help write assignments. Is this a common problem everywhere and how is your Uni managing it?

Yes it’s common. We’ve developed a new policy to attempt to deal with it. All the assessment as been overhauled to try and ask questions which are more difficult for AI to answer well. Much of the AI usage isn’t done well, however, so it is fairly easy to pick up (I am in a fairly niche field though). We’re probably going to move back to in person exams next year.

Not a uni but a college and yes it’s the first year we have really seen it. AI is fairly easy to spot in marking but we are looking at technology that spots it too

In my department cases of suspect use of AI are referred to the Academic misconduct process. The Programme Leaders meet with the student to discuss concerns and ask them explicitly. Then based on that and our own evaluation of the assignment, we are able to decide whether or not the student is guilty of academic misconduct. Another way to reduce AI use is authentic assessments.

Yes it is a problem but very hard to prove the use of AI or to do anything about it.

I think shifts to in person exams and more vivas are likely

Yes we're seeing increased use of AI in assignments - and as someone who is in a departmental academic offence adjudicator role it's taking up more and more of my time. Chat GPT and similar LLMs and text spinners feature prominently.

There's no institutional desire to return to unseen examinations - they're viewed as pedagogically poor learning indicators.There's talk of using more 'creative' forms of assessment (testing a students knowledge of Kantian Ethics via interpretive dance possibly).

I am hoping we may have a more creative approach next year as a department as our current workshy departmental academic officer rotates on to a new admin role. Unless we have a student who cops to it when asked, he refuses to deal with it as an academic misconduct offence. Thankfully almost all the ones I’ve suspected of AI usage have also been utter shite generally.

Yes, of course. The cat is out of the bag, the genie is out of the bottle, it's going nowhere.

We're being encouraged to allow its use within certain parameters.

It's very hard to prove usage, so even if you make it an academic offence it's hard to actually prove it to a degree that will stick in an appeal.

The software for checking isn't accurate either. I wrote a piece of text recently, all my own words, ran it through as a test and it came back with about 70% chance that AI had written it.

testing a students knowledge of Kantian Ethics via interpretive dance possibly

To be fair, DD would love that. Her imitation of an apple crumble boiling over in the oven has gone down in family legend, so I'm sure she could gesticulate her way through Kant.

My uni is trying not to be overly defensive about it. They are trying to work with it, but not produce prompt engineers and nothing else. I like it.

And as @Fiery30 says, we are pretty good with authentic assessments.

But one stand out example from last semester was a reflection piece. 500 words reflecting on the group work experience...the submission was OK until... 'as I am not human, I do not feel emotions, but if I was a human, I would feel...'

That was a particular highlight for me.

That's a brilliant story!