We're sorry. Something went wrong.
We are unable to fully display the content of this page.
The most likely cause of this is a content blocker on your computer or network.
If you continue to experience issues, please contact us at 202-466-1032 or firstname.lastname@example.org
Most colleges and universities started this fall without redefining existing plagiarism policies in light of AI, instead treating its use as a matter best left to instructors. In the absence of guidance or resources from their employers, many of us overworked and anxious instructors have fallen back on familiar solutions. The issues are acute for those of us who regularly teach essay writing, in particular. Some instructors have announced plans to scrap it all in order to keep the robots out of the classroom. Others, more comfortable using new technologies as pedagogical aids (often digital humanists and education experts, but also ed-tech consultants), have proposed teaching with AI, not against it. Speaking anecdotally, it seems as if most people I know are trying to split the difference, putting their heads together to draft AI policies for the classroom.
Given the options and the lack of guidance, it makes sense to have a policy. The problem with creating policies, however, is that they have to be enforced. This burden of enforcement can create a huge amount of extra work for instructors: running essays through plagiarism-detection software, or surveilling in-class exams rather than trusting students with a take-home. In addition to normalizing the routine policing of students, that extra work has other negative consequences for some of the most vulnerable. Plagiarism-detection software frequently gives false positives when the author is a non-native speaker, while in-class exams can disadvantage students experiencing food insecurity, fail to correlate with long-term content retention, and have the potential to violate the “reasonable accommodations” provision of the Americans With Disabilities Act.
The problem with creating policies is that they have to be enforced.
Rather than normalizing policing that most instructors are neither qualified nor supported to do — and that constructs punishments rather than moral accountability — we might instead consider what ChatGPT’s response to my query implied: What’s the human context for all this cheating? One possibility, of course, is that students are inclined to malfeasance. Perhaps they are contemptuous of learning in all its forms, and are looking only to secure the highest grade with the least amount of work. In that case, cheating would be the student’s fault in a pure sense, and the context in which cheating occurs would hardly matter. But none of that has ever been my experience of students. More often, contexts matter hugely.
First and foremost, for many students, is the economic context: 38 percent of college students work 30 hours a week or more. And those numbers don’t even take into account the hours of internships and other “opportunities” to work at no pay that keep attracting students in pursuit of something practical, called “experience,” that they seem to think colleges don’t otherwise offer them. One context in which people cheat, in other words, is one in which they’re strapped for time.
Meanwhile, even students who are shielded from economic worries may still find essay writing unduly difficult from a technical point of view. Only about 27 percent of K-12 students perform at the National Assessment of Educational Progress’s “proficient” level in writing. There may be a host of interconnected factors behind those low numbers — the abandonment of phonics-based reading instruction in the 1990s; the Bush-era No Child Left Behind Act, which emphasized standardized testing and pedagogies that “teach to the test,” often at the expense of creativity or critical thinking; the post-2015 Every Student Succeeds Act’s shift away from common state standards, very much including language-arts curricula; and of course the 2020-21 year of remote Covid instruction, with its unprecedented learning loss. Whatever the cause, it seems safe to say that another context in which people cheat is one where they’re being asked to do something they’re not prepared to do.
At the intersection of economics and preparation are families. Multigenerational households are on the rise in the United States, fueled by economic factors, and are more likely to affect the classroom due to demographic shifts in college populations, including more students from cultural backgrounds where such households are already common. When students say, for instance, that a grandparent died, they may be referring to a vastly more emotionally and economically disruptive experience than instructors with nuclear families on the brain will assume. A third context in which people cheat, intersecting with the other two, is one where they’re psychologically strained or emotionally overwhelmed.
Contexts aren’t causes, and in my experience teaching writing, there are plenty of students who are or become strapped for time, underprepared, or overwhelmed, who may flail and fail but don’t resort to cheating. Nonetheless, what makes cheating with ChatGPT different than many kinds of cheating that have long existed is, I think, the speed and ease by which it works to mitigate exactly the challenging contexts in which the current generation of students now works: by freeing up time, supplementing deficits in their preparation, and asking nothing emotionally. It’s an all-too-ready fix for any student whose circumstances are desperate enough to gamble against a punitive classroom policy. If that’s right, then the antidote to cheating is probably not plagiarism-detection software. Policing students punishes their actions but does nothing to mitigate, let alone acknowledge, the contexts that motivate them toward such actions in the first place.
The good news is that, unlike AI policies, this kind of abolitionist work has a lot of existing support, though it often goes by other names. Many existing learner-centered pedagogies have abolitionist effects, which can be used to address the problem of cheating by tackling its causes. Consider this nonexhaustive list of some well-tested pedagogical strategies that can be used to work toward abolishing cheating in essay writing:
- Overarching narratives. Syllabi and lectures that present, connect, repeat, and reinforce a big picture, helping students understand the relation of the parts to the whole and thereby what they’re supposed to be learning and why.
- Transparent design. Assignments that test students on content but also require them to reflect on how they’re learning and why, teaching them how to be better learners in the future.
- Labor-based grading. Assignments that promote learning through process and practice, rather than evaluation and outcomes, rewarding current effort rather than previous preparation.
- Pink Time. An exercise that requires students to skip class in order to reflect on personal motivation in the context of greater autonomy.
- Collaboration. Students who work in teams through multistep assignments, focusing on the work-flow, organization, delegation, and teamwork skills required for many 21st-century jobs.
- Accessibility auditing. Assignments designed to accommodate all students regardless of disability, making further accommodation probably unnecessary, evaluation more fair, and the classroom a model for inclusion as a standard.
All of those strategies emphasize process and practice, de-emphasize high-stakes evaluation and final products, and work to ensure greater equity in the classroom. A number of them are designed to increase students’ individual and collective capacities for reinforcing learning and undertaking evaluation, with the consequence that the work process that leads to and through essay writing can carry equal if not greater weight than the completed essay itself. If what students think matters is the finished essay and its grade, ChatGPT will always produce results faster and more easily than students can. If what they think matters is the multistep, collaborative process of writing and revising, well, that’s exactly what ChatGPT robs them of.
More effective than to police cheating would be to abolish it.
Does this really work? In my experience, yes, it really does. In the past several years, I’ve tried all of those strategies in different combinations with what have felt like fairly clear successes. I’ve seen remedial writers improve faster and further through low-stakes practice, and I’ve seen more motivated students take uncommon initiative in leading groups and organizing lessons. Class dynamics have gelled better, and, through the Pink Time exercise in particular, students have processed feelings of curiosity and strangeness when they were left to both learn and evaluate that learning for themselves. By designing assignments that anticipate access accommodations in advance, I haven’t had to scramble to arrange them along the way. There have been zero issues with plagiarism, and grade complaints have dropped precipitously. Meanwhile, the students who resisted those pedagogies the most were the ones used to getting A’s without having to reflect on their work process or create learning structures for themselves.
None of the pedagogies on the list above specifically seek to abolish cheating, but their use may mitigate the contexts that push students toward cheating. Of course, college instructors can’t stabilize the economy or prevent the death of grandparents; bad things will still happen and put pressures on students. But by policing cheating, instructors punish students for their contexts when we could — with arguably less effort, rather more good will, and quite possibly much greater success — employ pedagogies that absorb some of the shock those contexts create.
“Abolition requires that we change one thing,” writes the geographer and prison abolitionist Ruth Wilson Gilmore, “everything.” And changing everything cannot but be a process. If you want a world without cheating, one starting point, where instructors have quite a bit of authority to make immediate changes, is to create a classroom without cheating. As the pedagogical strategies listed above already suggest, tools to promote that kind of change are tested, supported, outlined, and otherwise extant. The change is already here if you want it.
“But that’s just not how I teach!” you may be saying. Well, thanks to ChatGPT, the way you teach also isn’t how you’re going to teach, because the conditions that have enabled that pedagogy are rapidly changing. Right now, therefore, is a pretty good time to pay attention to the university spaces in which abolitionist thinking is already being practiced, because that’s where you’ll find some of the most transformative change accomplished with the fewest top-down resources. Whichever end of the political spectrum you land on, there is widespread agreement that the possibilities for cheating with ChatGPT require instructors to teach differently than in the past. Meanwhile, all of the new policies and tools that have been developed to police cheating have consistently failed to make it stop.
Writing is a technical process, and a way of communicating information, but one of the contexts in which it happens is also, in my experience, a social and emotional one. Drafting this essay involved moments of frustration and other moments of satisfaction. I doubted myself and fretted, eventually finding my stride, losing it, and finding it again. As I wrote sentences, I imagined audiences and their reactions. I asked myself objective questions like “Is what this sentence says factually true?” but also human questions like “Is what this sentence says fair?” I had conversations with my editor and solicited feedback from my friends. The finished essay is the result of a process that operates in a human social context. I think that context is meaningful, and I think its meaning should be appreciated. But ChatGPT isn’t going to help students understand what that meaning is. Neither is calling the academic police.