Resources

Like so many of us, I’ve spent the past two years in a paralyzed panic over artificial intelligence’s effects on my classroom. I teach undergraduates, mainly gen ed philosophy courses, and writing has been a key component of all my courses. When ChatGPT hit the mainstream, it became a constantly looming presence, threatening to devour every part of teaching that I care about. I didn’t “wrestle” with it. Nothing so active and dignified. I went on an emotional roller coaster of ignoring it, freaking out, wishing it away, catastrophizing, and then ignoring it again.It didn’t work. AI was still there. I tried writing about it, but that just made me feel worse. And my writing was awful, page upon page of “Oh my god, the sky is falling.” Depressing, unhelpful – and bad writing. I trashed every single page.Some of my colleagues argue that we must incorporate this wonderful new tool into our teaching. We should encourage students to use AI for “basic” tasks like summarizing texts and outlining arguments, freeing them up for more advanced work. Others point out that summarizing and outlining are advanced tasks for many of our students since they don’t know how to do either, and that students need to first acquire skills like summarizing in order to later acquire more advanced skills. To make that learning possible, they argue, we need to build protective walls to keep AI out of our classes. Several want our Writing Center to ban Grammarly and its ilk.I agree with the second group that our students usually don’t summarize or outline well. And I agree that allowing students to outsource tasks they haven’t yet mastered to AI will make it harder for them to learn to read, write and, most importantly, to think critically. I’d love to operate in a sheltered space behind protective walls. But I don’t think the walls will hold.Hence my freaking out. But after two years, I have finally managed a few moment of calm thought, aided by James Lang’s wonderful blog post. I’ve come to the following key conclusions:AI-assisted writing isn’t going away. Damn it.We aren’t reliable AI detectors and we don’t have reliable automated AI detectors (although we can catch blatant and unskilled uses).If we continue to assign take-home essays, some of our students will use AI to write them. We won’t know how many or how much they will use it, and we won’t catch many of them.Take-home essays are important pedagogical tools, and I don’t as yet have any promising substitutes.My immediate task is to figure out how to navigate my classroom spaces with all this and my own teaching goals in mind. What do I want to prioritize, and what am I willing to sacrifice?It is tempting to prioritize not being duped. And making not being duped the priority has the clear advantage of producing simple action steps: No more take-home essays. Switch to lockdown browsers or old-school blue book exams.Following James Lang, I am not switching, at least not yet. This is because I think there are more important things at stake than minimizing the risk of cheating.As I listen to colleagues who are switching to in-class exams, I am thinking about why I’ve been avoiding them for my entire teaching career: They do not test what I want to teach.Switching from essay-writing to in-class exams requires moving from messy and open-ended discussion towards lectures. I don’t want to make that move. My students have enough lecture classes. They don’t need another one from me. But they do need what I am good at teaching. My students need a class that focuses on discussion and self-reflection, inviting them to engage each other and the materials and think through their own lives, actions, and values. I want to teach those classes, and then I want my assessments to provide opportunities for students to chew over things we’ve talked about and the views they’ve encountered in class, developing arguments, reflecting on their experience, pursuing thoughts and objections, and seeing where it all takes them. Take home essays do that.But assigning those essays leaves me wide open to cheating. So what do I do in my classes to reduce the risk?I include more low-stakes writing.I make the papers worth less and include plenty of scaffolding and in-class work on them.I grade a little differently, rewarding bland, generic, but correct writing less and messy and creative writing more.I add some quizzes – and I am experimenting with using AI to draft multiple choice questions.I keep an eye out for obvious AI misuse and I use the built-in detection software. But I try not to obsess about it, and I try to be OK with knowing that some students will get away with things they shouldn’t (this part is definitely a work in progress).Most importantly, I try to connect with my students and I try to convince them that I want to hear what they think, and that their opinions matter to me and to the world. I encourage them to draw on class discussions and their own experiences when they write, and I encourage them to say what AI cannot say because AI is not them.I’m also looking around for guidance from others. Reading a Chronicle of Higher Education newsletter, I just came across Kimberly Kirner’s writing assessments. She sets out to help her students develop their own voices, and she grades based on the students’ progress towards goals that they develop together. I plan to learn from Kirner and others like her over the summer and experiment with her assignments next semester.AI is here to stay and our students have access to it. It’s not the situation I would have chosen but it is what is in front of us. It will be on us as educators to guide students so that they can still develop as critical thinkers and writers. That work has many parts, and thankfully we don’t all have to do all of it. Despite the peptalks from the AI-optimists on my campus, I don’t see myself working with students to help them write better AI prompts, and I don’t yet see a good role for AI in my courses. But reading Kirner and Lang reminds me that there is important work here that I am suited for and that I care about: I can help students see that they and their voice matters and I can help them develop their voices and become better informed so that they can speak and write more effectively. Notes & BibliographyKimberly Kirner is Professor of Anthropology at California State University at Northridge.James Lang is Professor of Practice at the Kaneb Center for Teaching Excellence at the University of Notre Dame.

Without fail, every recent conversation that even remotely touches upon assessment leads to an increasingly intricate and technical discussion about how to prevent the cheating, presumed to be rampant, now that we’re all online. Should we use Lockdown Browser? Should we enable webcams? But what about smart phones? What proctoring services are available? Will time limits help? How many questions should we have in our question bank? What’s a good proportion of questions in the bank to questions drawn for the exam? Should we shuffle questions? Should we show only one question at a time? Should we allow students to see their incorrect answers right away? The questions concerning cheating proliferate—with each solution presenting further considerations and, often, new challenges. These conversations can’t help but remind me of one of the many theories for the pandemic “panic buying” of toilet paper: psychologists claimed that such hoarding was an attempt to retain or regain a sense of control over an uncontrollable situation, in which feelings of fear, disorientation, and overwhelm (in addition to actual physical danger) abound. We can’t control Covid-19 (yet), but we may be able to control how many toilet paper rolls we have stashed away in our homes. Perhaps teaching online feels similar to instructors: something that was thrust upon us, something most of us didn't want to do, something that most of us probably aren’t doing very well (either by default or by design). Trying to tamp down on cheating may be the academy’s version of toilet paper hoarding. When I’m part of these conversations, I keep James M. Lang’s Cheating Lessons in mind. (If you don’t have time for the whole book now, he wrote a series in The Chronicle of Higher Education that offers many of the same insights in shorter form.) Here are a few important considerations about cheating: • So far, we don’t have evidence that students are cheating now more than ever—though our worries about cheating have obviously increased. Even typical indicators of cheating, for instance, fast time on a test or an unusually high score, may be attributable to other factors. One colleague told me that some of her students were studying more at home, without all the distractions, and were more relaxed about taking tests—which led to better performance. • Most students in any given class won’t cheat (even if most people admit to having cheated at least once over the course of their lives; I know I have); this seems to be an especially important time to be viewing our students charitably, with positive regard, and avoiding a “deficit” lens. • Why do people cheat? Well, among other reasons, we know that certain conditions in a learning environment can incentivize cheating; one example is a single, high-stakes assessment for which students have been given little to no practice or preparation (e.g., one exam worth 50% of their course grade). To the extent that we can give students multiple opportunities to demonstrate knowledge or mastery of a skill, to the extent that we can offer them practice and timely feedback, and to the extent that we can help reduce their anxiety at this anxiety-inducing time, we will go a long way in creating conditions where they will not make the choice to cheat. • What is cheating anyway? We can’t assume students will know in the context of our specific class; even within the same field, the same department, these definitions and expectations may vary (just as there is no consistency across what we mean by different assignment types). Some students, like the first-generation population, are severely disadvantaged by this kind of “hidden curriculum.” The less that’s tacit, the more that’s explicit (or “transparent”)—on cheating and otherwise—the better for all students (but especially for underserved populations). • But okay, let’s assume, for a moment, that students know it’s cheating to look up answers online or in their books, and they do it anyway. One colleague of mine said, “Who cares? The point is I want them to learn. Maybe looking up the answers will help them learn.” What is our end goal here? How can we leverage or even embrace (perhaps newly) open testing situations? • Or, rather than focusing on the cheating that we defeatedly accept will occur—and trying to prevent the inevitable with complicated surveillance and other apparatuses—perhaps we might sidestep the issue entirely by designing our assessments to be as “cheat-proof” as possible. Questions of foundational knowledge are easy to look up; “higher-order” questions—application, analysis, evaluation, or creation—are not. (“Which of the following is the best definition of karma”? is easy to find in the textbook I use. “What would a Marxist interpretation of this [current news headline about religion]?” isn’t.) For those teaching large classes, those with heavy teaching loads, those without TAs, those stretched to capacity already, etc., these questions needn’t be asked only in short-answer or essay form, which can take way longer to grade. Multiple-choice questions can work here too, even if we don’t usually think of them as such. (A caveat: some folks find these kinds of multiple-choice questions harder to create.) I’m not convinced that we have to accept cheating as inevitable, even now. We, as instructors, have some control(!) over the learning environments, the testing conditions, and the assessments themselves. Let’s continue to use the influence we do have to encourage authentic, deep, transformative learning—not short-cuts or hacks.