Students and Generative AI

When the Turing Test was passed in November 2022 with the release of ChatGPT, things changed for teachers; especially teachers who rely on the essay as being their “gold standard” for assessment. Suddenly students can utilize generative AI to complete written work for them, leaving some teachers floundering.

AI detection tools like Turnitin or GPTZero are tempting to use. The teacher takes the studen’s written work, loads it into one of these detection tools, and the tool confirms or refutes the teacher’s perception that the written work may have been completed by AI. Easy, right?

There are actually a few problems in this scenario. We’ll go through them one at a time here.

They don’t work

It has been shown repeatedly in the empirical literature that AI detection apps fail, indeed research reveals that these detection tools remain unreliable (An & James, 2025; Moorhouse et al., 2023). Classroom climate can be quickly flushed by wrongly accusing a student of utilizing generative AI for an assignment.

The teacher-student relationship takes time to develop and it serves a powerful pedagogical purpose in the classroom. When the relationship is destroyed, it impacts not just the teacher and one student, it can have much larger impacts than that, and it would be a genuine shame for this pedagogical tool to be obliterated by a false result from an AI detector.

It’s an “Arms Race”

Villasenor (2022) stated that in the arms race between writing tools and detection tools “the AI writing tools will always be one step ahead of the tools to detect AI text” suggesting that as fast as the detection tools can catch up to the generative AI tools, the tool students use to write will also be moving ahead. Van Dis et al (2023) noted the same, stating that “such detection methods are likely to be circumvented by evolved AI technologies and clever prompts” (p. 225). This alone suggests that these tools will not yield the result that the sleuthing teacher is seeking.

One Step Ahead

It should also be noted that students who would choose to use AI to complete their writing for them will likely also use social media apps like TikTok to gain new techniques to conceal or “humanize” the text they are intending to submit. There are many content obfuscation techniques that a student may put into play if they are the type student who would undertake such an action.

How Prevalent is the Problem?

It’s difficult to get a gauge as to the actual numbers of students who will choose this method of cheating on their schoolwork. The companies who make these apps to detect academic. misconduct have impetus to make claims that are higher than what the literature would indicate the incidences of actual cheating usage, as they are selling a product. It’s in their best interest to make claims about the extremely high numbers of “gotcha” documents as a means of convincing potential customers that their product is valuable.

Should Students use Generative AI?

According to Weber-Wulff et al (2023), “the use of AI tools is not automatically unethical. On the contrary, as AI will permeate society and most professions in the near future, there is a need to discuss with students the benefits and limitations of AI tools, provide them with opportunities to expand their knowledge of such tools, and teach them how to use AI ethically and transparently” (p. 2). One of the basic beliefs teachers hold is that they are preparing students for the real world, and to that end, it will be important for teachers to make their peace that AI is here to stay, and in order to appropriately prepare students for a future that will include AI, it will fall to teachers to adjust the means of accomplishing assessment in ways that are beyond the essay. This is not to say that we must abandon the essay as an assessmentt mechanism; but it does demand some innovation and rethinking on the part of teachers everywhere.

So yes, students should use AI when it is appropriate to do so, and teaching students HOW to use AI ethically and appropriately is going to fall on the shoulders of teachers to do this.

So What Should a Teacher Do?

AI detectors don’t work, and a false accusation can destroy the teacher-student relationship. So, what should a teacher do if they suspect that a student has utilized generative AI to write a document that is part of the class assessment?

This is actually where you need to lean on the teacher-student relationship, and it can actually be an opportunity to further build that relationship, believe it or not.

But you’re going to have to play “Columbo” for a few minutes.

(Apologies to the younger generation who are not so familiar with the old Columbo movies. Columbo was a popular American mystery TV series where the title character, a seemingly bumbling but highly intelligent LAPD homicide detective, solves murders; basically seeming like he just couldn’t put the pieces of the mystery all together, when in reality he was a homicide detective who appears unassuming and disorganized to hide his sharp, observant mind.)

What I mean is that you’ll need to have a conversation with the student you suspect of having used generative AI to write their work, and you may have to lay it on a little bit thick. Maybe try something like this:

The Script

Teacher: So, I read your essay over the weekend, and wow!!! Has your writing ever improved this year!! When I compare what you turned in for this essay to the work you were writing in September (pick your date/time in the past) I am blown away.

(watch for signs of discomfort; fidgeting, facial redness, beads of sweat, aggression)

Teacher: Here’s the thing though, it’s my job to teach and assess the curriculum, and because your writing has improved so dramatically, it’s my job to make sure that you understand the outcomes on the [insert name of course you are teaching said student] program of studies. So, I’m going to ask you some clarifying questions to ensure that your comprehension of the curriculum is at the level that this writing would suggest that it is.

(continue to watch for signs of discomfort)

Teacher: When you wrote [insert phrase from student writing that seems unlikely that they actually wrote] what did you mean? How did you draw that conclusion? [Ask any question that occurs to you with respect to the writing they submitted.]  Just be Columbo. Be confused, don’t reveal your cards, and don’t make an accusation. 

At this point you will be precipitously close to having the student confess.

Ask another clarifying question. If the student actually wrote the work, they should have no problem answering your questions, and you should be able to actually mean your compliments of their writing if they are able to answer your questions.

If the student cannot answer your questions, but will not confess, provide them with a sheet of paper and a pen or a pencil and ask them to write a summary paragraph that would allow someone who has never heard of [insert topic of the essay here] before to understand the fundamental premises of the essay. 

Again at this point, you’re on the verge of the truth here, and no accusation has been made to the student. 

Factor what they write down on that sheet of paper into their grade.

Scale this to your Entire Class

You may want to consider scaling this summary task to your entire class. Have every student complete this task in the moments after they submit their essays to you for grading. It matters not whether they submit their work to you in a LMS like Google Classroom, or if they print their work and hand it in. Ask every student in your class to take out a single sheet of paper and a pen or a pencil. They need to write a summary of their essay in class without the essay or a computing device. Just a pen and paper summary of what they just handed in to you. 

If you build the accountability in to your system, they’ll choose to use AI for someone else’s essay. You won’t be the target of this misbehaviour for long.

Villasenor, J. (2023). How ChatGPT Can Improve Education, Not Threaten it. [EB/OL] [2023-05-14]. Available online at: https://www.scientficamerican.com/article/ how-chatgpt-can~improve-education-not-threaten-it/