Faculty in higher education are navigating a genuine tension right now: how do you design for authentic learning when a student can produce a polished written response in seconds? The answer isn’t to ban the tools—it’s to rethink what you’re actually trying to assess. A recent Panopto webinar explored how video can help navigate that balance. Dr. Derek Bruff—Associate Director at UVA’s Center for Teaching Excellence and co-author of the forthcoming Norton Guide to AI-Aware Teaching—joined Ian Paice, Panopto’s Chief Customer Officer, to share six strategies faculty are using right now. The throughline across every strategy they shared: video captures the process of learning, not just the product—and that distinction is what AI can’t replicate.
Oral exams are back—and video makes them scalable
Oral exams had largely fallen out of favor in higher education, but faculty are rediscovering them as a richer form of assessment. Research shows students find them more demanding and more personal than written ones, and something shifts in how they prepare: more group study, more thinking out loud, more focus on actually understanding rather than reproducing.
The video format makes this practical at any scale. A short recorded submission works asynchronously across disciplines and can double as a portfolio artifact. One education policy professor at UVA realized his oral component was more than an AI deterrent—talking about policy with other people is itself a transferable skill, and he rebuilt the assignment around that. In a music cognition course, each student-created explanation video became a node in a class-wide concept map. In broadcasting, oral assessments became professional demo reels. The format shift matters—but so does who students are talking to.
Assign real audiences to raise the stakes of student work
Dr. Bruff has a useful frame for why so many assignments fall flat: the rhetorical triangle. In a traditional assignment, a student explains something to the person who already knows it best. There’s no real communicative stake; just performance for evaluation. Shift the audience to someone who genuinely knows less, and something changes. Students become the authority. They have something real to say.
Video is one of the most direct ways to make that shift. In a public speaking course, Ian described, students filmed speeches in front of a live audience—no edits, real people present—and the recordings became the basis for self-assessment and peer feedback. At a vocational training institution, trade students recorded hands-on assessments in real conditions, with instructors watching live, and then reviewed their own footage to spot gaps themselves. Changing the audience changes the stakes. Changing the conversation around AI changes the behavior.
Make student AI use visible with simple documentation practices
More than half of students use AI even when it’s banned, often simply because no one has talked to them about it openly. Research behind the TILT framework (Transparency in Learning and Teaching) is consistent: when faculty are explicit about the purpose, task, and criteria of an assignment, students engage more genuinely. That clarity matters even more when AI is part of the picture.
One approach Dr. Bruff shared uses an “AI consultation form”: students document how they used AI, what they were trying to accomplish, and what they learned. It makes AI use visible and reflective, rather than hidden. Paice described a similar experiment at a UK institution: shared visibility into student processes helped faculty spot unproductive habits early, and good approaches spread through the class on their own. Once students are documenting how they work, you can ask them to learn something from it.
Use video to help students reflect on their own learning process
Once students start documenting their process, you can ask them to learn something from it. In one computer science study, students recorded screencasts of programming tasks—once with AI, once without—then watched the footage back. What they noticed surprised them: patterns in their own thinking they’d never seen before, moments where they reached for help too quickly, places where the struggle turned out to be useful.
Dr. Bruff frames the best AI-integrated assignments as doing three things: building domain knowledge, developing AI fluency, and creating space for reflection. One colleague designed an assignment where students pick a persuasive task they’ve never pulled off, work with AI as a planning partner, and reflect on what the collaboration made possible: not just whether it helped, but how and at what cost. Process reflection works outside of class. Inside the classroom, the same principle applies differently.
Flip your classroom to protect time for higher-order thinking
The flipped classroom is an older idea that feels newly relevant. Move information transfer outside of class through video, and bring the harder work—applying, questioning, making sense of things together—into the room where students have each other and their instructor present.
An English professor at UVA moved all his out-of-class writing into the classroom. Students turn off Wi-Fi, dim the lights, put their word processors in focus mode, and write for 45 minutes together. It sounds strict, but every student across two semesters said they preferred it to writing at home. Ian Paice shared a biology professor who embedded pre-class videos with quizzes, then used Panopto analytics to identify where students struggled. Reserving class time for hard thinking only works if students arrive prepared to do it—and the same design mindset applies to AI tools. You said: update H2s to be sentence case
Build custom AI tools shaped around your course—before students build their own
Most students don’t wait for faculty to recommend an AI tool—they find one on their own. And general-purpose chatbots are designed to answer questions completely, not to teach. Dr. Bruff’s phrase “good fences make good chatbots” points to the alternative: tools designed to guide students toward understanding rather than supply the answer, grounded in course content and shaped to behave like a tutor rather than a search engine.
Faculty doing this well aren’t just enabling a tool; they’re designing an experience. One biochemistry professor built a custom tutor bot loaded with his own course materials, so students got explanations aligned with how he was actually teaching the subject. Another team wrote over 1,200 words of instructions to shape how their bot interacted with students. That care in the design is the whole point.
Each of these strategies works because it shifts the focus from what a student produces to how they think. Video is the thread connecting them—it makes the process visible in a way that a written submission never could. When faculty design assignments around that visibility, they’re not just making AI harder to misuse. They’re creating conditions for the kind of learning that actually sticks. Tools like Panopto give faculty the infrastructure to make that shift practical at any scale.
