I’ve written previously about my personal approach to artificial intelligence (AI) as a classroom teacher, and I’ll probably write about it again because that approach is evolving almost as fast as AI itself. Every day, it seems, brings some new challenges related to my students’ use of AI. I suspect many of my colleagues are experiencing the same thing.

That’s why it’s important for institutions to develop AI policies that are rational and sustainable. Right now, it appears to me that the three primary approaches to AI at the institutional level are (1) ignore it and hope it goes away; (2) reject it as uniformly harmful; and (3) fully embrace it and expect everyone on campus to do the same—none of which strikes me as particularly useful, long term. A sustainable AI policy must acknowledge reality while prioritizing student learning and preserving academic freedom and faculty autonomy.  

Clearly, AI is here to stay, in some form or other. Is its current popularity overblown—a bubble that, like the dot-com bubble in the 1990s, will ultimately burst? Perhaps. But I notice that the internet hasn’t gone away—far from it—and I assume the same will be true of AI. It will likely be a major part of the economic and social landscape throughout our students’ lives, so it behooves us to help them learn to navigate it.

And I have no problem with that. I believe one of the primary goals of colleges and universities moving forward must be to teach students how to use AI well and ethically. Clearly, certain disciplines, such as computer science, are uniquely equipped for that task; others, including those that are highly data-driven, such as the sciences and social sciences, can undoubtedly assist.

At the same time, we must also focus on helping students develop their human skills—qualities machines can simulate but not really replicate—such as critical thinking, creative writing, public speaking, and interpersonal communication. Other disciplines—namely, the humanities—are better equipped for that task.

A sustainable AI policy must prioritize those skills. To the extent that AI becomes a crutch for students, a way for them to outsource their thinking or offload difficult, time-consuming tasks such as writing, it must be eschewed. Some cognitive abilities can be developed only through deep thinking and attempting to organize one’s thoughts on paper. Teaching students to do that has always been one of the main purposes of a “liberal arts” education, and it must remain so even in the age of AI.

No doubt this will be difficult to enforce. Most students don’t want to think deeply because it’s hard, and they don’t care much about writing well, which is also hard. That hasn’t changed over my 40-plus years in the classroom. Today’s students just have more tools for avoiding the hard things. Many will passive-aggressively rebel if not allowed to use AI in every situation. Some will cheat—and, as we all know by now, there’s no reliable way to prove AI cheating.

But that doesn’t mean faculty shouldn’t try to limit students’ AI use where we believe such restrictions are appropriate. It comes down to a mindset. What are we trying to teach them? What do we value? Permissiveness is not the answer because it sends the message that we don’t really care about things like writing and critical thinking, either. We’re just hypocrites. So we must at least try to hold the line.

Fortunately, there are ways to make our courses more “AI resilient”—not AI proof—such as carefully crafting our assignments to make them more difficult simply to farm out to AI and revising our grading standards to prioritize qualities like originality, creativity, personal connection, and voice over mere correctness.

Thus, AI should certainly be allowed, even taught, in some courses but not allowed, or at least severely limited, in others. But who decides? That’s where the final characteristic of a sustainable policy comes in: it must be driven by the faculty, not the administration.  

At many institutions, AI policy has so far been very much top-down, with the upper administration essentially telling deans and department chairs, who then tell faculty, to “get on board” with AI. That’s the wrong approach, and it will ultimately be counterproductive.

Think students are passive-aggressive? There has never been a more passive-aggressive group of people in the history of the world than university faculty, especially the tenured variety. If they don’t want to do something, they just won’t do it. They’ll find every excuse, every workaround, every delaying tactic possible. If upper-level administrators want faculty to take a productive approach to AI, they must leave the details to the faculty.

This is, of course, in keeping with the well-established role of faculty as the custodians of the curriculum, as enshrined in the American Association of University Professors’ 1940 Statement on Academic Freedom. It is also a staple of shared governance. The people who decide what should be taught in a given discipline are the discipline experts: the faculty.

That includes AI. How much or how little should students be encouraged or even allowed to use AI in a given course? That should be primarily up to the person teaching that course, within broad guidelines established by faculty curriculum committees—not the president or the provost, much less the board. Obviously, even within a given discipline, not all faculty members will agree. But they can hash out their differences in committee and department meetings and ultimately reach some sort of compromise. That’s how shared governance is supposed to work.

At the university level, then, a sustainable AI policy would, first, recognize that AI is likely to play a major role in our students’ lives; second, prioritize student learning over every other consideration; and, third, empower faculty to determine how best to facilitate that learning as it relates to AI. Such an approach would be flexible enough to address the ever-changing nature of technology while preserving the institution’s core mission and maintaining the traditional balance of power between faculty and administration.  

Follow Rob Jenkins on X.