AI and Academic Honesty

[Jason]

With the rise of AI, how can you trust that academic papers are real? In fact, paper writing will have to be monitored, but probably there are no plans to do this, my guess.

 

[Arantor]

Proctoring is hard, expensive and often doesn't prove squat. (Source? I used to work in a directly adjacent field, providing e-learning platforms to major UK universities.)

This is a major topic of discussion at present but honestly they're not taking it the way you think. There are the traditionalists who take the view that AI is a scourge and should be eliminated, and honestly it's not that hard to tell if a paper is AI-backed if you regularly interact with the student, because you'll get a feel for vocabulary etc.

There are people out there who get pissy that students use spell checkers. And even tools like Grammerly to improve their ability to communicate. How *dare* they use the tools out there to improve their output! I've even seen people demand things on paper as 'proof' of not cheating (but all it proves that unless it's done under vigilation conditions, that students will just use whatever tools they were going to use, then copy it down by hand before hand-in)

Then there are the technologists who assume that a) students are going to use this, therefore b) the test is not 'does the student use tools' but 'is the student using the tools effectively and not as a crutch'. This is a much more interesting question: if a student does use ChatGPT does that make them a fraud? The answer is, surprisingly, 'not necessarily'. Prompt engineering *is* a thing, as is the necessary ability to edit and refine the content to say what is going to be said.

This also of course is more problematic when you consider how much time in academia spent writing research grant papers and not actually doing research itself, for which ChatGPT is a much more appropriate tool for the job.

 

[Jason]

This is a major topic of discussion at present but honestly they're not taking it the way you think. There are the traditionalists who take the view that AI is a scourge and should be eliminated, and honestly it's not that hard to tell if a paper is AI-backed if you regularly interact with the student, because you'll get a feel for vocabulary etc.

Maybe, but can teachers really take the chance?

 

[Arantor]

I even answered what the other teachers are doing, where they're assuming students are using this anyway... it's only the ones who look backwards who are taking this approach at the moment.

 

[Jason]

I even answered what the other teachers are doing, where they're assuming students are using this anyway... it's only the ones who look backwards who are taking this approach at the moment.

A big problem would be really good students who could probably write spectacular content and the teachers know that, so in that case, the student would be lazy, use some AI to fool people.

 

[Darth Cognus]

If a student is that spectacular and then resorts to AI I have a feeling that sharp change in quality will be noticed. The thing about AI is that it writes in a specific way and because it does not truly understand what it's saying, it will more than likely spout nonsense.

In other words you're a bloody genius if you naturally write in AI structure, manage to have passing grades, and then when AI hits off start using it to fill in the gaps in a way that isn't genuinely noticed. Can't rule out a lazy reviewer though I guess. But I don't think at that point they'd bother much about this topic.

 

[Jason]

If a student is that spectacular and then resorts to AI I have a feeling that sharp change in quality will be noticed. The thing about AI is that it writes in a specific way and because it does not truly understand what it's saying, it will more than likely spout nonsense.

In other words you're a bloody genius if you naturally write in AI structure, manage to have passing grades, and then when AI hits off start using it to fill in the gaps in a way that isn't genuinely noticed. Can't rule out a lazy reviewer though I guess. But I don't think at that point they'd bother much about this topic.

Yeah, professors and teachers need to be aware of how AI comes across when writing so they can detect it.

 

[Arantor]

in a way that isn't genuinely noticed.

This, according to the folks I know that actually have this challenge. In practice it's actually similar to the human detection of plagiarism - yes there's an automated check via something like TurnItIn but honestly, the lecturers can usually tell by mismatches in tone and language. For the folks who just wholesale reject AI involvement, it's already being shown not to be that hard to get it right reasonably consistently.

And the ones who embrace it, they're not even looking at that anyway, but that's my original point...