It is almost a cliché to say that generative AI can be both a challenge and an opportunity in education. Here’s a story that captures this duality perfectly, showcasing two different approaches to how generative AI can be used in educational contexts, one ethically problematic (to say the least) and the other quite creative and constructive.

This story comes from my colleague, Dr. Jim Dunnigan, who shared it on our AI in Education Slack channel. Jim is a professor in our EdD program and one of the leading exponents in our college of the intelligent use of generative AI in doctoral education.

As it happens, Jim had encountered a case of AI plagiarism in one of his classes. The telltale signs were there: unusually formal language, formulaic answers with patterned phrases, and a conspicuous lack of personal references. The submission stood out from others in structure and length, missing specific details that couldn’t be provided by a simple prompt.

But Jim also knew that there was no real way of “proving” what his instincts honed through years of experience told him. The track record of plagiarism detectors is mixed, at best.

Jim realized he would need to have a conversation with the student. And this is where Jim’s approach took an interesting turn. Instead of immediately confronting the student, he turned to AI himself – not to catch the student out (as some faculty have tried to do, unsuccessfully), but rather to prepare for the conversation.

Using Claude, an AI language model, Jim essentially role-played the upcoming interaction. He asked Claude to develop a strategy for the conversation and then simulate how it might unfold. Jim practiced different approaches, received feedback on his communication strategies, and refined his technique – all before the actual meeting.

Jim shared the entire conversation he had with Claude—here are some highlights.

Claude suggested starting with a neutral, non-accusatory opening and explaining the purpose of the meeting. It advised Jim to ask open-ended questions about the student’s writing process before presenting his observations.

Jim practiced phrases like, “Can you tell me more about how you structured your paper?” Claude, playing the role of the student, responded with explanations that allowed Jim to practice active listening and guide the conversation towards academic integrity. What added a layer of authenticity to the simulation was Claude’s ability to capture and describe the physical and emotional responses of the simulated student. For instance, Claude would preface the student’s responses with descriptions like “hesitates for a moment, then speaks in a slightly defensive tone” or “shifts uncomfortably in the seat, avoiding eye contact.” These details helped Jim anticipate and prepare for the non-verbal cues he might encounter in the actual conversation.

The AI even reminded Jim to discuss proper and improper uses of AI in academic work, suggesting he say something like, “It’s perfectly fine to use AI to help with your writing – we all do that now. The key is you cannot use it to write entire sections for you.”

This preparation allowed Jim to approach the real conversation with confidence and clarity. When he finally sat down with the student, the discussion, in Jim’s words, “came off beautifully.”

---

What is important to note here is the stark contrast between the student’s actions and those of Jim. The student had used AI to shortcut learning, claiming generated work as their own. Jim, on the other hand, used AI as a tool for professional development, enhancing his ability to handle a delicate situation effectively.

---

This is similar to work that Rezwana Islam and I presented at the 2024 SITE conference, in a session titled On Having Difficult Conversations: Leveraging the Social Nature of Generative AI. In that presentation we had explored the potential of generative AI as a tool for educators to practice and refine their skills in managing difficult conversations. By simulating various scenarios – such as addressing student tardiness, managing conflicts with teachers, and communicating with struggling international students – AI can provide a safe, controlled environment for educators to experiment with different approaches and receive immediate feedback.

This approach offers several advantages: it allows for repeated practice without real-world consequences, provides consistent scenarios for skill development, and can be customized to address specific challenges faced by educators. Moreover, it demonstrates an innovative use of AI in education that goes beyond content generation, focusing instead on enhancing interpersonal skills and professional development.

The work that Rezwana and I presented at SITE last year, was exploratory in nature. Jim showed us how this could actually play out in the real world.

His story serves as a reminder of the potential of AI in education when used thoughtfully. It’s not about finding quick answers or avoiding cognitive effort. Instead, it’s about exploring possibilities, refining our skills, and ultimately improving our ability to connect with and guide our students.