Who's Going To Uni?

[scottsdad]

4 AI cases now. I think I've cracked it. A few students had a very strange way of formatting their answers. I put the exam question with a little specificity into ChatGPT - exactly the same format comes out. 

But I'm being nice. I'm storing the paperwork until January. In years gone by I'd have sent the "You're caught, boyo" email out before Christmas. I'm going soft. they can receive it on January 9th. 

ETA - the answers are mince. The same item listed as an advantage and a disadvantage. No detail whatsoever. 

Edited December 20, 2023 by scottsdad

[GHF-23]

1 hour ago, scottsdad said:

4 AI cases now. I think I've cracked it. A few students had a very strange way of formatting their answers. I put the exam question with a little specificity into ChatGPT - exactly the same format comes out. 

But I'm being nice. I'm storing the paperwork until January. In years gone by I'd have sent the "You're caught, boyo" email out before Christmas. I'm going soft. they can receive it on January 9th. 

ETA - the answers are mince. The same item listed as an advantage and a disadvantage. No detail whatsoever. 

Sorry sure this has been covered before but have you always had this problem even pre-AI text gen? As a very good boy who never copied an essay (although I did always have to pull an all nighter to finish almost everything, including my dissertation (first) and my Higher English teacher once barged into a Modern Studies Classroom to accuse me of plagiarising because I had read and referenced the introduction of The House with the Green Shutters and a couple of articles about the kailyard school in my big book report assignment) I didn't think I knew anyone who did c.2011-15 but maybe all the b*****ds were at it

[scottsdad]

3 minutes ago, GHF-23 said:

Sorry sure this has been covered before but have you always had this problem even pre-AI text gen? As a very good boy who never copied an essay (although I did always have to pull an all nighter to finish almost everything, including my dissertation (first) and my Higher English teacher once barged into a Modern Studies Classroom to accuse me of plagiarising because I had read and referenced the introduction of The House with the Green Shutters and a couple of articles about the kailyard school in my big book report assignment) I didn't think I knew anyone who did c.2011-15 but maybe all the b*****ds were at it

Plagiarism is a wee bit different and miles easier to detect. With AI systems the trouble is that you can ask the same question a dozen times and it'll give you a dozen similar (but not identical) answers. Hard to detect (hence why I'm quite chuffed to have cracked this one)

As much as the tools are good (eg turnitin) there's no substitute for experience and a complete and utter mindset of suspicion. 

[mathematics]

1 hour ago, scottsdad said:

a complete and utter mindset of suspicion. 

And if you don’t have one of these, after marking for a year you soon will have.

[Central Belt Caley]

3 hours ago, scottsdad said:

4 AI cases now. I think I've cracked it. A few students had a very strange way of formatting their answers. I put the exam question with a little specificity into ChatGPT - exactly the same format comes out. 

But I'm being nice. I'm storing the paperwork until January. In years gone by I'd have sent the "You're caught, boyo" email out before Christmas. I'm going soft. they can receive it on January 9th. 

ETA - the answers are mince. The same item listed as an advantage and a disadvantage. No detail whatsoever. 

You’ve probably mentioned before but what sort of punishment do students get for stuff like this? Do they get off lightly if they own upto it? 

[scottsdad]

31 minutes ago, Central Belt Caley said:

You’ve probably mentioned before but what sort of punishment do students get for stuff like this? Do they get off lightly if they own upto it? 

First offence, resit the module.

After that it gets very steep! Up tonexpulsion.

[Richey Edwards]

Passed this term despite public speaking anxiety getting the better of me during a verbal presentation which I eventually made a rip-roaring c**t of.

In the end, despite that debacle, no great harm was done and I still got an overall pass on the strength of my written work. I can look back on it and laugh, take the lessons from it and move on. 

I eventually will have another verbal presentation to do for another module later in the year during my final semester. I will try to do better in that. 

[scottsdad]

11 hours ago, Richey Edwards said:

Passed this term despite public speaking anxiety getting the better of me during a verbal presentation which I eventually made a rip-roaring c**t of.

In the end, despite that debacle, no great harm was done and I still got an overall pass on the strength of my written work. I can look back on it and laugh, take the lessons from it and move on. 

I eventually will have another verbal presentation to do for another module later in the year during my final semester. I will try to do better in that. 

A few of my colleagues ask for students to make videos and submit these, to take the sting out of standing in front of a class and talking. 

I had presentations in a class of mine years ago. Quickly dumped it - some students get really anxious. For me it isn't really assessing if they have met the learning outcomes or not. I get that for some subjects it's important (a mate of mine lectures in history, and as such a lot of the class will go on to be history teachers). 

[vikingTON]

23 hours ago, scottsdad said:

4 AI cases now. I think I've cracked it. A few students had a very strange way of formatting their answers. I put the exam question with a little specificity into ChatGPT - exactly the same format comes out. 

But I'm being nice. I'm storing the paperwork until January. In years gone by I'd have sent the "You're caught, boyo" email out before Christmas. I'm going soft. they can receive it on January 9th. 

ETA - the answers are mince. The same item listed as an advantage and a disadvantage. No detail whatsoever. 

How do you plan to present this as demonstrable evidence of misconduct, if a student contests the charge? In most cases you'd expect them to just admit responsibility and take the L, but I'm not 100% convinced that this method would stand up to scrutiny if a student pushed their case. 

That's not a criticism of your efforts as there's no other way of trying to flag this issue right now; but 'similar format' and 'mince answers' are not quite smoking guns that a student disciplinary process can rely on. 

[mathematics]

21 minutes ago, virginton said:

In most cases you'd expect them to just admit responsibility and take the L, but I'm not 100% convinced that this method would stand up to scrutiny if a student pushed their case. 

It’s amazing what a towel and a little water can do to loosen the tongue.

[scottsdad]

32 minutes ago, virginton said:

How do you plan to present this as demonstrable evidence of misconduct, if a student contests the charge? In most cases you'd expect them to just admit responsibility and take the L, but I'm not 100% convinced that this method would stand up to scrutiny if a student pushed their case. 

That's not a criticism of your efforts as there's no other way of trying to flag this issue right now; but 'similar format' and 'mince answers' are not quite smoking guns that a student disciplinary process can rely on. 

I get what you're saying. 

Firstly, misconduct cases are decided on the balance of probability rather than reasonable doubt. 

So, I put in my case. I show the ChatGPT answer and the student answers and highlight the similarities. I then hand it on to the misconduct team. As the instigator of the case my involvement ends there. The student will be invited to a hearing and will have access to my evidence. They can explain their side to the committee, who will then decide guilt or not. 

For me this is looking clear cut. The structure of the answer is the exact same as the AI one. The wording is very, very similar. Especially in the introductory and concluding paragraphs. So the question for the committee is - is this a coincidence? The student just happened to write their answer in this particular way? With 8 bullet point answers (in the exact same order) with such similar wording, lack of detail, etc? Or did the student use an AI system. 

[scottsdad]

Let me give an (anonymised) example I came across this morning. The question asks about durability of materials. The student answer started with "Certainly! Let's look at...." which is a very odd way to begin an answer. 

I went to ChatGPT and typed in "Can you give me a description of the durability of...." and filled in the rest of the exam question. The answer I got back was exactly the same. "Certainly! Let's look at..."

[VincentGuerin]

2 minutes ago, scottsdad said:

Let me give an (anonymised) example I came across this morning. The question asks about durability of materials. The student answer started with "Certainly! Let's look at...." which is a very odd way to begin an answer.

[Richey Edwards]

15 minutes ago, scottsdad said:

Let me give an (anonymised) example I came across this morning. The question asks about durability of materials. The student answer started with "Certainly! Let's look at...." which is a very odd way to begin an answer. 

I went to ChatGPT and typed in "Can you give me a description of the durability of...." and filled in the rest of the exam question. The answer I got back was exactly the same. "Certainly! Let's look at..."

Fucking hell, if you're going to cheat at least don't begin your answer with "Certainly!". 

[Mon eh Ton]

9 hours ago, scottsdad said:

For me this is looking clear cut.

I've not read much of the thread so don't know which uni you're at, but we've stopped reporting seemingly obvious cases because the conduct assessors don't see anything as clear cut and it's all just a waste of our time. Unless the students leave the "Certainly! Let's look..." stuff in - although even then the penalties are laughable. Frustrating knowing they're getting away with chancing it, but most of the answers tend to be crap anyway.

[scottsdad]

ChatGPT cases I referred - all found guilty.

[101]

1 hour ago, scottsdad said:

ChatGPT cases I referred - all found guilty.

Hopefully when they fessed up they replied "Certainly! I used AI"

[Richey Edwards]

Currently writing my independent research paper. It's a pain in the ass.

[scottsdad]

9 hours ago, Richey Edwards said:

Currently writing my independent research paper. It's a pain in the ass.

Did you change specialism to proctology?

[Richey Edwards]

Just now, scottsdad said:

Did you change specialism to proctology?

Not yet.