AI and Student Agency

It's not just about us

In 2023, most of the emerging advice on AI was directed to teachers—and mostly on how to curtail AI use. I couldn’t find anything that talked to students in a straightforward and honest way about the harms and benefits of AI. So, Tim Laquintano and I decided to write something that respected students’ choices in using AI, something that was candid about the capabilities of AI and the fallibility of AI detectors. We wrote up an honest letter to our students, Dear Students: Should you use AI for writing?

Here’s how we open it:

Dear Students,

Should you use AI for writing?

Well, to start with, we’re not going to answer the question for you. The point of this essay is to help you--and your instructor--answer that question. AI is everywhere in writing right now, and there are no easy answers about how or whether to use it.

We talk about how AI works, why it was suddenly on everyone’s minds, and how it can help and harm their learning processes. I took this same approach when I wrote the chapter “Navigating AI as an Author” for the 5^th Edition of W. W. Norton’s Everyone’s an Author.

I think our Dear Students document is still useful. But thankfully, now there are a lot more resources directed towards students, including a guide by and for Duke students (my favorite because it’s by students!), an engaging and concise video series by J.T. Bushnell at University of Oregon, a highly-produced guide (in English and Spanish) by the AAC&U and Elon University, and a guide from University of Michigan that includes some helpful questions students can ask themselves. For environmental considerations, I love both Jon Ippolito’s “What uses more?” tool and Susan Ray’s Custom GPT Carbon Footprint Chatbot, which help students calculate the environmental impact of their lives, including their uses of AI.

These tools don’t tell students what to do about AI—instead, they equip students with information to help them make decisions. They treat students as agents in their own educational lives. This is crucial. Students are navigating difficult decisions about AI all the time. They’re being aggressively marketed to by AI companies and “homework help” platforms, they’re observing and judging their peers’ uses of AI, and they’re subjected to a wide variety of AI policies in their courses. They have to sort through all this and figure out what to do.

Students make decisions about AI

Last week, I visited Seton Hill University to give a workshop for students and faculty together. This is my favorite workshop format because it gets everyone talking together. (Thanks to Patrick Berry for coming up with the idea for my visit to Syracuse!) I generally begin by explaining why faculty are anxious about AI (everyone is cheating their way through college, we’ve been told!) and how students are experiencing aggressive marketing, mixed feelings, and many pressures regarding AI. Using Mentimeter as an anonymous survey platform, I ask both students and teachers to tell the other group what they really want to say about AI.

One positive thing about generative AI is that it’s an opportunity for faculty and students to learn together. That is, if faculty can let go of our customary authoritative position. We don’t know more about AI than students do; in many cases, we know less. But we do know about our subjects and other contexts for AI. Coming to the table with different expertise regarding AI, we can learn from each other through honest dialogue and exploration.

The Seton Hill students really showed up. It was beautiful. They said a lot of what I’ve heard from students at Pitt, Syracuse, and elsewhere. Around 10% of the ~80 students said they refused to use AI. The rest said they used it for writing, reading, research, emails, and fun stuff. They told us: AI was helpful for them; it stripped them of creative thinking; it wasn’t always cheating; it made their lives easier; it was terrible for the environment; it isn’t always evil; they didn’t like being accused of AI just because they were good writers; and too many of their peers got away with it. In other words, they have complex feelings about AI and they don’t all agree.

The Seton Hill students responded—as students always do—as people with agency about AI. Whether they’d shared their thoughts on AI with their teachers or not, they’d thought about it. They’d considered course policies, environmental impacts, technological affordances, and pressures on their time. Like students everywhere, most of them had come to the decision to use AI at least some of the time. Of course, not every student thoughtfully considers every angle when deciding about AI. They may be unduly influenced by peers and Big Tech or they may be miscalculating the actual carbon footprint of a chatbot query. But the point is that they are all agents in that decision.

The limits of teacher authority

These thoughtful responses from students are why I am increasingly frustrated at the framing around AI approaches that center solely on teachers’ agency. The principled rhetoric around refusing or embracing AI in courses assumes the choice is up to us: We have thought about what it means to use AI, and, through our course policies, we will let our own decisions about AI govern yours. It’s a fundamentally authoritative stance: I know what’s good for you. And the problem is that students no longer buy this stance.

For better or for worse, faculty can no longer rely on our position as authorities. Student uses of AI have revealed this fact: they sometimes cut corners for work they find to be unaligned with their interests or professional trajectory, regardless of course policies. We’ve also seen a resistance to faculty authority through political pushback from students.

Given a lack of assumed authority, we can’t just run through our syllabi and assignments according to our own priorities. We must justify assignments, point out the benefits of gen-eds, and make a case for college in general. I don’t think this is necessarily a bad thing. It forces us to re-examine why we teach what we teach. But perhaps more exciting--and challenging--is that we are forced to acknowledge students as agents, navigating our classes as thinking, writing, voting individuals who use various technologies, have diverse backgrounds, and believe different things about the world.

Which is why I think we should be talking to students more about AI. But more importantly, we should be listening to them about their concerns and interests in using AI. I am continually surprised at how few pedagogical resources about AI include student voices. We should be working with students to help them make decisions about AI that align with their own commitments and values.

Mark Marino provides a great example of working with students in his Analog Sandwich course design, where students lean into AI, then go fully analog, and finally make their own decisions about how to integrate AI in their writing for the course. Sydney Sullivan also accounts for student agency when she asks: “what would happen if we stop telling students how to use AI?” and provides a roadmap for collaborating with students on AI policy. The tools and guides I linked above are also good ways to start a conversation with students.

Teachers matter in a classroom. We can shape discourse and policy and share what we think and know. However, we can’t compel students to believe what we do about AI. Instead, we should learn together with students, listen to them, and guide them—in the best way we can—to exercise their agency wisely.