Co-Creating Institutional AI Policy with Students: A Student-Centered, Open Pedagogy Approach

From Concern to Course Design

In late December 2022, just weeks after ChatGPT entered public conversation, I was grappling with the same questions many educators were asking. How would students use this? How would faculty respond? How quickly would generative artificial intelligence (Gen AI) become framed as an academic integrity problem? At College Unbound, those questions felt immediate. My colleague, Autumm Caines, had already identified an early instance of possible student use, and as Director of Faculty Development and Innovation, I was trying to think through how to respond carefully without setting negative precedents too quickly.

As Autumm and I talked, we realized students had likely already begun experimenting with these tools and that curiosity was needed. Instead of treating student exploration only as a problem to manage, we looked at what it could mean to work through this moment with students rather than about them. That shift led quickly to a practical response: a new eight-week course on AI and education in which students would help develop an institutional AI usage policy for faculty and students. A second session would later pilot and revise that policy before it moved into faculty review and formal approval.

My default approach to course design is to use open pedagogy and agency woven throughout the course. Rather than asking students to complete a disposable assignment about AI, the course centered on institution-facing work with authentic consequences. In that sense, the case reflects open education through student agency, collaborative inquiry, and work that moved beyond the classroom into institutional practice ( DeRosa & Jhangiana, 2017; Wiley & Hilton, 2018).

College Unbound’s context helped make that approach possible. The college serves primarily adult learners and emphasizes cohort-based and project-based learning, along with a robust credit for prior learning program. Students begin projects early and carry them forward across their studies. Asking them to engage a real institutional problem matched the kind of applied, consequential work the college already values.

Building the Inquiry

In late December 2022, just three weeks after the release of ChatGPT, my Provost greenlighted my proposal for a course for January 2023 that focused on AI and Education. The course was timely; it was also intense and a bit nerve-wracking. The tools were new, public understanding was unstable, and even the basic terms of the conversation were still forming. The course had to do several things at once: introduce students to the tools, create space to experiment with them, examine their implications for education, and turn those experiences into policy language that could be useful to the institution.

From the start, students wanted to understand what these tools were, what they were good at, and what kinds of questions they raised. Because College Unbound serves adult learners, students brought a complex range of lived perspectives into the course, more than typically seen in a traditionally aged classroom. They engaged AI not only as students, but also as parents, multilingual learners, and professionals working in education, social work, healthcare, and other fields. Those perspectives widened the inquiry. They were not simply asking whether AI could help complete schoolwork. They were also asking how it might affect communication, labor, trust, access, and judgment across different parts of their lives.

To begin, everyone in the course, including me, completed weekly logs using ChatGPT.^[1] Students were asked to engage in at least 10 prompts each week, often around a theme or task, and then share what they noticed.^[2] These logs gave us a common base of experience. They helped students see what felt useful, what felt awkward, what seemed misleading, and what raised ethical or practical concerns.

That early work also surfaced confusion. Students sometimes struggled to know what to ask, how to evaluate responses, or where the boundaries of acceptable use might be. Instead of treating that uncertainty as evidence that students should be kept at a distance from the policy conversation, the course treated it as a reason to involve them more fully. Over time, students’ questions became more nuanced, and many of the concerns they raised in those early weeks were concerns I would later hear repeatedly from faculty: authorship, overreliance, transparency, privacy, fairness, and the difficulty of making decisions while the tools were still changing.

The next assignment asked students to draft their own AI usage policies. Some wrote broadly; others focused on particular contexts such as workplace use. This step moved the inquiry from experimentation to position taking. Students had to ask what they believed about use, disclosure, boundaries, and responsibility before trying to shape policy for anyone else.

From Individual Positions to Collective Policy

Once students had drafted individual policies, we began asking what an AI usage policy should look like for College Unbound. Each student brought a distinct set of concerns, hopes, and assumptions into that discussion. The first collective draft was, in many ways, a (cinematic) Frankensteinian policy: stitched together from multiple positions and not yet fully coherent. We needed to reconcile different positions about lenient and strict expectations, what percentage of an assignment could include AI content, and how thorough are citations (e.g. should usage include chatlog printouts?). We sought a workable starting place that made these differences visible.

As the draft took shape, the questions became more specific. How detailed should the policy be? How much control should it attempt to exert? What kinds of AI use required citation or disclosure? How should expectations apply to students and faculty, particularly if transparency and fairness were shared values? What responsibilities did the college have around privacy and tool use? How should the policy address authorship, accountability, and support without collapsing into suspicion?

These questions intersect with technical, ethical, and practical tensions. The class was writing policy while the tools themselves were evolving, while norms were unclear, and while both underreaction and overcontrol carried risks. Too little guidance could leave students and faculty uncertain. Too much control could produce a brittle policy disconnected from actual practice. There were also questions about whether institutional language could remain fair to different learners, including adult and multilingual students whose experiences with writing, labor, and technology might not fit standard assumptions.

Disagreement became part of the work. When students differed, we tried to understand why. Sometimes the difference came from confusion, sometimes from stronger concern, and sometimes from a personal or professional context that changed how a student viewed AI use. At times, students were often aligned on values but differed on how explicitly those values should be written into the policy.

One strategy that helped us move forward was asking whether an issue was a must-have or a would-like-to-have and an explanation of why or what they meant by that. That distinction gave students a way to identify what was essential and where they were more flexible. I also occasionally added context about faculty concerns or institutional constraints so students could better understand the environment their policy would enter. Even then, the center of gravity stayed with student deliberation. They were doing the work of translating values into language and language into institutional guidance.

Testing, Revision, and Approval

By the end of the first eight-week session, the class had produced a complete draft of an institutional AI usage policy. Rather than treating that draft as final, I offered a second eight-week session in which students piloted the policy in other courses.^[3] They then reflected on the policy, their usage, and how it impacted their learning. Faculty were informed that students would be testing how the policy functioned with actual assignments.

That pilot stage gave the class something more than speculation. Students could see where the policy was too restrictive, where it was too vague, where it clarified expectations, and where it created new confusion. The policy then went through another round of revision based on those experiences. The sequence of inquiry, drafting, pilot testing, and revision gave the document more grounding than a one-time classroom exercise would have allowed.

After that second session, the policy moved beyond the classroom. It was shared with faculty across the institution in a Google document for feedback. Faculty comments focused largely on clarity, consistency, and symmetry between expectations for students and expectations for faculty. Students and I reviewed that feedback, revised the policy again, and then submitted it to the curriculum committee. The committee raised a few additional questions; those questions were addressed, and the policy was approved as institutional policy (See final draft).

What This Case Offers Other Institutions

The goal of this process was to produce a viable policy with students. This case shows how pedagogy can become a site of policy formation under conditions of uncertainty. Students were substantive contributors to the institution’s sense-making process rather than being treated as an audience for institutional decisions made elsewhere. They were substantive contributors to the institution’s sense-making process.

That is where open pedagogy was especially valuable and why I have found it to be a valuable approach in teaching and learning. The central assignment asked students to do work with visible consequences beyond the course itself. In a moment when generative AI was forcing educators to ask harder questions about what counts as meaningful academic work, this approach invited students into a form of learning grounded in responsibility, negotiation, and public consequence. The assignment did not eliminate uncertainty around AI, but it gave students and the institution a way to work through that uncertainty together.

The case also suggests that AI policy benefits from processes that are iterative, stakeholder-informed, and connected to actual teaching and learning. Students engaged the tools directly, reflected on their own practices, drafted and revised policies, tested them in context, and helped shape institutional adoption. That sequence aligns closely with a pedagogy-first approach to policy.

Not every institution can or should build policy through a single course. Still, the broader lesson travels. When institutions face emerging technologies, students can contribute more than feedback after the fact. They can help name the questions, test the assumptions, and shape the policies that will affect their learning.

AI Usage Statement

For this essay, I used GenAI to help interview me and externalize the different parts of this project. I then used AI to help draft a detailed outline. After writing and editing, I further used AI as an editor for further organizational and flow support. All work in this essay reflects my thoughts and insights and was reviewed line by line.

References

DeRosa, R., & Jhangiani, R. (2017). Open pedagogy. In E. Mays (Ed.), A guide to making open textbooks with students. Rebus Community.

Wiley, D., & Hilton, J. L., III. (2018). Defining OER-enabled pedagogy. International Review of Research in Open and Distributed Learning, 19(4).