Recently, I took a fully online, asynchronous course at an accredited institution. About halfway through the course, I received a notice of disenrollment. I called the dean’s office and was told that I had violated the honor code related to using generative AI to complete an assignment.
While I do use generative AI for some things, I’m super careful to cite it when I do, and I didn’t remember using AI for any of my work in that course. The office advised me to contact my professor.
When I did, I received an email response: I turned you in for cheating because your paper was too well researched, too sophisticated, and too thorough for a beginning-level class. I tried sharing my “evidence” of how I did not use AI, but my professor just followed up with, We’ll follow the procedure that academic affairs has laid out for plagiarism.
I was gobsmacked. The assignment had asked us to explore a problem of practice in our lives, so I chose one deeply connected to my professional work: the legal and ethical implications of artificial intelligence in education. For more than two decades, my career has required me to think carefully about the intersection educators have with curriculum, technology and student data, including the past five years serving as a state lead on student data privacy. My 15-page paper examined issues I routinely wrestle with professionally: what school staff and students may overlook when using AI, the legal consequences of those choices, and emerging and pending cases that may reshape how schools approach AI. I researched, developed, and wrote the paper precisely because these questions matter to my work and because I understand the ethical implications of misrepresenting someone or something else’s work as my own.
What made the experience particularly troubling, however, was not simply that my work was questioned. It was how quickly a conclusion was reached without a conversation about how I had produced it. My professor knew little about my professional background, my experience with the subject matter, or the process I had used to research and write the paper. There was no conversation in which I was asked to explain my argument, discuss my sources, describe my writing process, or respond to specific concerns about the paper. In a course centered on learning, that absence of relationship and dialogue struck me as especially consequential: human interaction could have provided meaningful evidence to prevent this.
What made the experience particularly troubling was not that my work was questioned. It was how quickly a conclusion was reached without a conversation.
A week later, I attended my virtual hearing on the matter. My professor began by explaining areas where he “knew” that my paper was generative AI: I had lengthy citations, the grammar was impeccable, and the writing was too sophisticated for someone in an entry-level course. When it was my turn to speak, introduced myself and explained to the panel my background. I gave them a copy of my CV and shared my process for writing my paper, which included screen shots of my writing history, and pulled up the articles that I had annotated over the years related to topic. To defend the claim that my writing was “too sophisticated,” I shared that the topic was one I have presented on more than five times and taught about in my school law class. I also shared some additional evidence, including a video of me talking about this same content for our state professional association.
Eventually, the panel and the professor withdrew their claims, and I was permitted to continue with the work in the class.
But I was left feeling uneasy. How do I go back to class after that? What do I say to the professor? What does the professor say to me? I scheduled a quick meeting with him just to break the ice: I approached the conversation as an educator rather than as an angry student. I told him I wished he had sent me an e-mail before turning me in, that if he had we might’ve been able to clear the misunderstanding up relatively easily.
Students Before Suspects
Ultimately, this experience became about something much bigger than AI use. It became a lesson about relationships. If educators—whether elementary, secondary, or college-level, virtual or in-person—do not take the time to know their students, uncertainty can turn into suspicion before a conversation ever has the chance to begin. Knowing a student does not mean ignoring evidence or excusing academic dishonesty. It means having enough context to ask questions before drawing conclusions.
Relationships, then, are not simply something we build to make students feel welcome. They are part of how we teach well, how we interpret what we see, and how we respond when something does not seem right.
My experience happened in a graduate course, but the lesson extends well beyond higher education. Imagine a middle or high school student receiving an accusation of AI misuse from a teacher who knows little about the student or how that student works. A student may not have the confidence, experience, or even the language to challenge the conclusion or explain what happened. As AI makes it increasingly difficult to draw conclusions about authorship from a finished product alone, knowing our students and knowing their work becomes even more important.
With that in mind, educators can make two kinds of connections before drawing conclusions about a student’s work.
Get to Know the Person
Build relationships before you need them. In K–12 classrooms, teachers often have opportunities to develop relationships naturally. Online and hybrid environments may require us to be more intentional about creating them. We do not have to set elaborate goals around relationship-building; small, consistent interactions can give us a much better understanding of the student behind the submission.
For example:
- Create a discussion board or online space where students introduce themselves and share something they care about and respond to each student individually.
- When possible, connect instruction to interests, experiences, or goals students have shared.
- Hold optional office hours where students can stop in to ask questions or simply talk.
- Send individual check-in messages before there is a problem to address.
- Instead of treating each interaction as isolated, intentionally circle back to previous conversations. A simple “How did that presentation go?” communicates that the earlier conversation mattered.
None of these practices proves whether a student used AI appropriately. But they can establish something that becomes especially valuable when questions arise: context.
Get to Know the Work
Knowing the student is only half of the equation. Educators should also become familiar with how students think, write, revise, and solve problems. A single finished product tells a teacher what was submitted; the process behind it tells us much more about learning and authorship.
Before we reach for detection tools or disciplinary procedures, perhaps we should reach for curiosity.
If a piece of work raises concerns, start by reviewing the student’s previous work. Does this submission differ from the student’s established writing, reasoning, or performance? If something seems unusual, begin with a conversation rather than a conclusion. A teacher might say, “Maria, I noticed this assignment felt different from some of your earlier work, and I wanted to talk with you about how you approached it.” Then listen.
We can also design assignments that make student thinking more visible. Ask students to explain why they selected evidence, describe how they developed an argument, talk through how they solved a problem, or revise part of an assignment during a conference. These practices are useful not simply for identifying inappropriate AI use; they are good teaching practices because they give us evidence of learning that a final product alone cannot provide.
When AI use is permitted, make that process visible too. Rather than asking only whether a student used AI, ask students to document how they used it:
- What prompt did you use and what did the AI generate?
- Which suggestions did you keep?
- Which suggestions did you reject?
- How did you revise or build upon the output?
Finally, collect evidence of the student’s process throughout larger assignments instead of waiting for the finished product. Notes, outlines, source selections, drafts, reflections, conferences, revision histories, and other checkpoints create a record of how thinking develops over time. They also create opportunities for feedback while learning is still happening.
Get Curious
Before we reach for detection tools or disciplinary procedures, perhaps we should reach for curiosity. I was not cleared because an AI detector was proven wrong. I was cleared because a hearing board finally took the time to understand who I was and how I had completed the work. Imagine how many similar conversations could begin and end differently if we started there instead.
Relationships will of course not eliminate academic dishonesty, nor should they prevent educators from addressing legitimate concerns about student work. But relationships create the context and trust necessary to investigate those concerns fairly. And knowing a student’s work gives us something more meaningful than a detector score: evidence of how that student thinks and learns.
AI has undoubtedly changed education. It has introduced new challenges, new ethical questions, and new responsibilities for both educators and students. But it has not changed one fundamental truth: students are still people before they are submissions, and educators are still mentors before they are investigators.
