Concerned About AI and Academic Integrity? We Have Ideas for You

In any conversation with teachers recently, three of the most likely conversation topics are the shocking levels of student apathy, pleas for texts that students might show interest in, and rampant concern about AI and academic integrity.

Really, these all boil down to the same problem: our students are increasingly disengaged and unwilling to read, write, or think for themselves, which leads us all to turn in desperation to the Internet: them for a bot that will write and think for them, us for anything that will capture their interest.

It’s hardly shocking that we have cause for concern about AI and academic integrity. After all, cheating has likely existed as long as school has, and the Internet has made it easier than ever to find, share, and, now, produce answers rather than doing the work to solve problems independently.

The increasing availability of generative AI tools like ChatGPT, however, has felt like an especially devastating blow to English teachers. We know our students will attempt to cheat, and we’ve all honed our strategies for detecting it. But what makes the conflict between AI and academic integrity so exceptionally frustrating is that use of AI is difficult to detect (and harder to prove), and tech wizards assure us that the quality of AI writing will only improve in the months and years to come. But if math teachers can make it work with the ubiquity of calculators, surely we English teachers can adapt as well.

AI isn’t going away and will grow in sophistication, which means we will continue to face issues about AI and academic integrity, and students will find new ways to use AI to complete assignments without detection. Moreover, students’ future employers will encourage the use of AI for its time-saving benefits, so while teaching students how to use AI isn’t our job (yet), neither ignoring it nor banning it is in anyone’s best interest. We have to accept what we cannot control and find ways to move forward.

But at the same time, there are still things we can do in the ongoing battle between AI and academic integrity. We’ve learned so much from engaging with colleagues around the country about their experiences and strategies, and Steph’s role teaching AP Seminar at a private institution has given her opportunities to try them out for herself.

We’ve rounded up a wide variety of strategies for reconciling AI and academic integrity in your classroom and holding students accountable for their learning, and we hope there’s at least something here that you find helpful.

Strategies for Navigating AI and Academic Integrity

AI and Academic Integrity Strategy #1: Set Up Your Assignments Intentionally

While we always want to assume the best of our students and start by giving them the benefit of the doubt, it’s wise to create our assignments with the strong likelihood of cheating in mind (and wise teachers have been doing this for years—after all, cheating isn’t new, only this particular method of cheating). This approach may not entirely prevent students from using AI to cheat, but it does make it more challenging for students to earn top scores on the assignment without doing the work themselves.

This is, by far, the most common advice we’ve heard from teachers around the country and a strategy we relied on heavily in AP Literature, even before ChatGPT.

Requiring students to write with pen or pencil on paper, without the use of any technological devices, while you watch them, is the most effective way to hold them accountable for doing their own thinking and writing.

If your school uses a program like GoGuardian that allows you to view students’ screens, you can monitor students’ AI use for a typed assignment, though it’s hard to monitor students’ screens and cell phone use at the same time.

This doesn’t, however, mean all your writing assignments have to be completed in class, by hand. If your goal is to assess students’ ability to construct an assignment or structure an essay, tasks AI is capable of passably carrying out, then you’ll want to limit students’ access to AI. If, however, you’re looking to assess students’ deep understanding of a particular text, topic, or idea, it’s unlikely AI will produce the level of depth or specificity you’re looking for, at least not without very savvy prompting (a worthy skill in and of itself, but not one many of our students have mastered).

We’ve discussed this approach in more detail in another post, but the more detailed and specific your requirements are, the less likely AI will be to generate the exact response you’re looking for. Referencing specific page numbers or quotations, requiring students to follow a specific format, asking students to explain an interpretation you have provided—all of these strategies make it harder for students to get an A without doing the work themselves.

The more modern the text, the fewer resources there are online for your students (or AI) to find. This makes it more likely students will have to formulate at least some of their own ideas about a text.

When you include questions that require students to describe a personal experience, discuss their personal reactions to a text, or apply the ideas in a text to their own lives, it’s easier to tell when they have asked a computer to do the work for them (and, in our experience, many students will just omit this part of the question, which prevents them from earning full credit).

While it’s not always practical to do regular student presentations, they can be a valuable tool in evaluating students’ familiarity with their own ideas. When students have to present orally, they often struggle to explain ideas they didn’t come up with themselves, so a presentation paired with a written component can serve as a “check” on students’ true understanding of the content.

AI and Academic Integrity Strategy #2: Set Up Your Class Structures with AI in Mind

There are also a couple things you can do from the very beginning as you set up your course and gradebook.

Because it has begun incorporating performance tasks completed during the school year into some of its courses, the College Board has developed a detailed acceptable use policy (see, specifically, the AP Capstone Policy) for teachers and students. The guide acknowledges that students are likely to use AI and that AI use can be appropriate and helpful, detailing in each step of the writing process what does and does not count as acceptable use.

Creating a similar policy that you distribute with your syllabus and other course materials can be valuable because it makes your expectations clear from the outset while acknowledging that there are valid reasons to use these tools when used ethically and appropriately.

Weight assignments for which you can control students’ environment more heavily than assignments you anticipate could be more easily generated by AI. If you incorporate a personal response or presentation component, consider weighting it heavily enough that students cannot earn an A on the full assignment without doing well on the personal or presentation part. Weight the quality of students’ arguments higher than anything else in the rubric (after all, ChatGPT tends to be a generalist). These are all ways that you can, perhaps not deter cheating, but minimize the impact of that cheating on students’ grades.

AI and Academic Integrity Strategy #3: Choose Your Battles

Ultimately, we can’t keep our students from using AI. They’re going to use it. And the world is changing: students hear constantly that they can and should use AI. We have to think about what skills an adult in our modern world needs to have when tools like AI exist.

After all, most adults rely heavily on tools like Grammarly, Google, and calculators to accomplish tasks in their personal and professional lives. Generative AI tools are rapidly becoming staples in the marketing and content creation world.

Unless we’re willing to drive ourselves crazy by policing every assignment (which we absolutely shouldn’t do), we have to prioritize those battles that are most important to us. Maybe grammar needs to be worth fewer points on our rubrics. Maybe homework assignments need to be worth less in our gradebooks. 

Save your AI-detective energy for the big-ticket assignments.

AI and Academic Integrity Strategy #4: Use Resources for Detecting AI Use

Thankfully, there are resources you can use to flag possible (and particularly blatant) AI use. However, and this is important to know, there are not foolproof methods for detecting AI use equivalent to Turnitin.com for flagging plagiarism. While there are tools available, they are widely known to be unreliable, and we’d feel uncomfortable relying on them in a teacher-said-student-said confrontation with a parent.  

One of the College Board’s generative AI policies is to incorporate multiple checkpoints throughout the writing process. Teachers meet with individual students to discuss their research process, the resources they’ve found, and the reasoning behind their essay outlines. Students who can discuss these things with some detail pass the checkpoint; students who cannot are asked to go back and revise.

It’s not perfect, and we have limited time, but finding opportunities to make our students talk to us about their work is one way to see that they’re doing at least some of the thinking.

Another widely used strategy for deterring (or at least reducing) AI use is requiring students to complete their assignments in Google Docs (or other Google apps) and give the teacher editor access. By doing so, students give you access to their revision history, which allows you to see their essay’s construction over time. It allows you to see if other students have edited the document and if large copy-pastes have taken place.

Knowing that your teacher can see your revision history is enough to deter our more honest students, and it provides the opportunity to have a conversation with other students. Even when they insist the work is their own, these conversations remind students of our expectations and let them know we are watching.

In addition to AI detectors, there are extensions that work with Google Docs to reconstruct a student’s writing process. After scouring the teacher online community groups, Steph started using Revision History to check AP Seminar work (Draftback was the other commonly recommended option).

Revision History tells you how long students spent actively writing in a document, identifies any large copy-pastes, and allows you to watch back a video reconstruction of the student typing, one keystroke at a time. It’s easy to see at a glance which students have only copy-pasted citations and quotes, still spending hours writing and rewriting; it’s equally easy to see students who have dropped enormous chunks of text into the document within only a few minutes.

How to Initiate Conversations About AI and Academic Integrity

One thing that was nice about regular old plagiarism is that you had clear evidence to present to the student, parents, and administrators.

With AI use, however, there’s no surefire way to prove it without literally watching the student copy-and-paste from ChatGPT into their Google Doc. There are many ways to flag a document as potentially having used AI, but we can’t say for sure, and when it’s us versus a student who insists they’ve done their own work, there’s a limit to what we can do.

Teaching AP Seminar online, Steph has found the best approach to be to raise a concern with the students without directly accusing them of cheating. This is the email she sends when she sees a concern:

This non-confrontational approach has worked well. Several students have replied with honesty, and we’ve worked together to figure out which parts of the paper they need to revise. Several students have shared the additional document they used to draft the paper. Several students have had a (dubious) story about the Notes app and typing directly into sites like Grammarly. In some cases, letting students know we are being vigilant and taking the issue seriously is the best we can do.

Ultimately, when it comes to AI and academic integrity, the best thing we can do is take a deep breath and focus on what we can control. It’s hard for us, as teachers, to accept that we can’t control everything, but we can do our best to create assignments and an environment where students know we will always be pushing them to do their own thinking.

We’re as new to AI classroom issues as many of you, so we’d love to hear more about what you’re doing to navigate this new challenge: reach out to us at [email protected] or DM us on Instagram @threeheads.works. 

If you’re interested in our suggestion about providing detailed directions but not entirely sure what that looks like, we’ve created a character analysis assignment utilizing our 5C paragraph structure that can be modified for use with any short story or novel. We’d love to hear if it helps your student produce more original work!