What We Heard at AAC&U’s Conference on AI and Higher Education
Last week, Lori Alvin, Sarah Cochran, Courtney Firman-Watkins, and Magdelyn Helwig attended AAC&U’s Conference on AI and Higher Education, where Furman presented three sessions. Courtney presented “AI Supporting Autonomy and Ownership for Instructors of a Common Curriculum: From Pilot to Scale.” Magdelyn presented “V is for Viable: Moving from AI Academic Integrity Policies to AI Literacy Integration.” She also presented “From Adopting AI to Studying AI: A Chatbot and Streamlined IRB Model for Faculty-Led Classroom Research” on behalf of Stephanie Freis.
On the final day, Lori, Sarah, and Magdelyn sat down to compare notes. We recorded the conversation, used AI to help turn it into a first draft of this article, and then edited the draft ourselves. Rest assured, there was a human (or two) in the loop.
The first thing we noticed was how differently each of us experienced the same conference. Depending on our session choices, one of us spent three days among English faculty and writing program administrators, another found herself primarily among faculty developers, and yet another sat at tables with provosts and an education representative from the National Geographic Society. We also noticed that institutions are in very different places when it comes to their approach to AI. Some are just starting these conversations, and others have been at it for four or five years. We came away feeling that Furman is ahead of the game in many ways. A campus-wide AI platform, shared AI literacies and competencies, assignment-level policy tools, and a summer faculty institute put us further along than many of the institutions we heard from.
Takeaway 1: The most important AI questions are about what makes learning human
José Antonio Bowen’s plenary asked whether we are asking the right questions about human thinking and AI, and that question carried through the whole conference. The most compelling work we saw didn’t begin with tools. It began with an institution or an instructor deciding which parts of learning are non-negotiable and should stay human, and then asking how AI might support those parts rather than replace them.
Speakers kept returning to the same analogy. If a machine lifts the weights for you, you don’t build any muscle. The point of the exercise is the struggle, not the weight. Learning works the same way. The value of an assignment lies mostly in the process of working through confusion, drafting, making mistakes, and revising, not in the finished product a student turns in. When AI shortcuts that process, the product may look fine while the learning never happens. That makes process the place to look when we decide where AI belongs in an assignment.
The answers vary by discipline. An instructor in a creative field might decide that the vision behind a work is what students must own. An instructor in a data-driven field might decide that interpretation is where students have to do their own thinking. Either way, the foundations still matter. Students who don’t understand a discipline can’t direct AI well within it, and they can’t tell when it has gotten something wrong. That’s why reflection came up so often. When students reflect on how they used AI and what it did, they learn about the tool, about working with AI, and about their own thinking.
For Furman, this points toward the curriculum. Miami University in Ohio has asked each department to identify a few courses where AI literacy belongs, in whatever form fits the field, from hands-on use to ethical debate. Our own AI literacies and competencies were designed with this kind of departmental work in mind.
Takeaway 2: Faculty work beyond the classroom is the next frontier
Nearly every session we attended focused on teaching or policy. Very few addressed how faculty might use AI in their own research, how that work should count in evaluation and tenure, or what IRBs should expect. Agentic AI is also raising new questions. These tools can act inside a learning management system on behalf of an instructor or a student, which could save real time but could also let a student hand off an entire course. Campuses will need safe places to experiment with tools like these before deciding what limits they want. Furman’s streamlined IRB model for classroom research gives us a head start, and AISTF plans to take up research and IRB questions next.
Takeaway 3: Good conversations about AI make room for grief
Many speakers described grief as a recurring feeling about AI on their campuses. We think that is worth taking seriously. Faculty who feel loss deserve acknowledgment, not a push toward enthusiasm. At the same time, productive conversation depends on people with moral objections and people with pragmatic aims making room for each other’s starting points. One way to build that kind of community is to share what we are actually trying in our courses. Many institutions now keep public repositories of AI-related assignments, and we would love for Furman to have one. If you have an assignment to share or would like to help build the repository, contact Magdelyn Helwig ([email protected]). For more resources, visit furman.edu/ai.
Acknowledgment: Claude Opus 5.5 wrote the initial draft based on our recorded conversation. Sarah and Magdelyn directed several major revisions, and Lori conducted a final review.