Last Updated: Winter/Spring 2023

Key Ideas:

  • Assignments that emphasize the writing process, rather than just the final product, will discourage students from using AI tools to do their work for them.
  • Process-oriented assignments also tend to be more engaging and impactful for students.
  • Framing your writing assignments as extensions of the class discourse challenges students to respond to the course content with new ideas of their own (something AI generally can’t accomplish).
  • Making revision integral to the assignment helps students critically examine and improve their writing process, while making it harder to take AI-assisted shortcuts.
  • Reflective and metacognitive writing assignments make students’ learning visible to them and to you; it also holds students accountable for the intellectual work of your course.

See assignment prompts that incorporate one or more of these principles.

At the time of this writing (winter 2023), the AI writing tool ChatGPT has recently launched, with several competitors soon to follow, and it’s clear that artificial intelligence will have a massive impact on the way we write, both inside and outside academia, for the foreseeable future.  As AI continues to evolve, so will our teaching practices, but one immediate concern for many instructors is the possibility that students may use ChatGPT and similar text-generation tools to avoid the intellectual work prescribed by many of our writing assignments.  

While the current generation of AI tools can’t produce particularly insightful or effective academic writing, that doesn’t mean students won’t try to substitute AI-written work for their own.  Furthermore, savvy users have already found more subtle ways to use ChatGPT to make their work as writers easier, such as using it to produce revisable first drafts or to shortcut the research process by identifying and summarizing a variety of sources on a given topic.  Whether these practices constitute academic misconduct or not is largely up to individual instructors to decide, but if you would personally consider them inappropriate for your courses, it’s worth considering how you might design new writing assignments or reconfigure old ones to make them less susceptible to digital assistance.

Fortunately, many of the existing best practices for designing writing assignments can also make them more difficult to complete with AI tools.  More importantly, though, they can also discourage students from turning to AI in the first place by placing greater value on the writing process, rather than just its product.  This helps to make the learning process more transparent and makes assignments more engaging, accessible, and impactful for students who complete them faithfully.

Making Writing an Extension of the Class Discourse

The best writing assignments will often ask students to engage not just the subject matter or source material of the course, but the discourse around that material that develops within the class itself.  A prompt, for example, might explicitly require students to respond to ideas that have come up in class discussion, or to apply concepts or themes that you’ve examined in class to a new text or subject.  However you frame it, the key to this approach is asking students to contribute something new to the discussion that explicitly builds on ideas they’ve encountered in your course.  In other words, ask students to make their writing an extension of the class itself, rather than an exercise that just happens to address the same basic material. This distinction can be subtle, but here are a few sample prompts that use this approach effectively.

This might not seem like a revelatory practice–after all, don’t all essay assignments require students to engage core ideas from the course?  But there’s an essential difference between simply engaging ideas from a course and responding to them effectively in conversation.  Engagement simply requires a student to understand the material and react to it.  Responding requires them to have a working understanding of the discourse around the material — who has said what, where previous speakers have agreed, disagreed, complemented and contradicted each other, and, most crucially, what remains to be said.  

As scholars, we all understand this implicitly; we review and cite other scholars because we want readers to see exactly where and how we’ve built upon the ideas that have come before, and we craft our arguments in ways that actively invite future scholars to respond to them.  Students, however, especially students who are new to college-level writing, tend not to think of their writing this way until they’re explicitly taught to do so.  High school classes and high-stakes standardized exams generally train students to see writing as a series of hurdles which they must complete for the sole purpose of demonstrating that they know the course material and can explain it in their own words.  Consequently, students often approach their writing exactly the way an AI would: by seeking to remix other people’s ideas in a way that fulfills the prompt, whether or not they say anything particularly new or interesting along the way.

Framing assignments as a way to build on the class discourse requires students to take a more active role through their writing, and in the course itself.  Once they come to understand that their writing will be an extension of the conversation taking place in the classroom, they start to realize that the more they participate in (or at least actively pay attention to) that conversation, the easier it becomes to engage it in writing.  They also become more invested in their writing, as they start to see the essay as a distillation of ideas that grew out of real interactions with real people, not a detached set of musings composed in isolation and shouted into the void (or the cloud).  

The writing produced this way is often messy, over-energetic, unfocused, underdeveloped, or otherwise heavily in need of revision. But it also tends to have a life and a voice that’s distinctive to the course, the term, the class, and the student all at once, and AI tools cannot match this distinctiveness. An AI can learn in the sense that it can gradually produce more refined output, but it can’t understand the ideas it synthesizes well enough to add anything distinctive and relevant to them. It can’t think, and that’s exactly what a good writing assignment will require students to do, above all else. Thus, any assignment that requires students to think — and to articulate their thinking clearly — in response to the ideas they encounter will resist AI’s intervention.

Creating Space for Revision

Most instructors encourage their students to make at least some effort to revise their writing.  We may implore them to start early, to consult the writing center, to read their essays out loud or to a friend, to form peer review groups, and so on. But it’s another thing entirely to make revision an integral and visible part of an assignment. Doing this not only places direct value on revision, but it also allows students to see and understand how revision works and why it’s so crucial to good writing and good thinking.

It’s important at this point to articulate a distinction that might seem obvious to any experienced writer, but is often revelatory to students: revision is not the same as editing. Editing (in this context, anyway) means examining and improving the presentation of one’s ideas–the grammar, the phrasing, the formatting, etc. Revision, on the other hand, means examining and improving the ideas themselves–fundamentally re-envisioning one’s conclusions and the tapestry of sources, responses, counter-responses, and epiphanies that lead to them. Most students don’t fully understand this distinction or, if they do, are not sure how to apply it to their own writing. They need to be taught how to revise, and this makes it hugely beneficial to them when revision becomes an integral part of an essay assignment.  When an instructor and a well-crafted assignment guide them through the revision process, many students discover for the first time what they’re really capable of as writers.

Granted, extended revision can be difficult to integrate into classes that are not explicitly designed to teach writing. Many instructors simply don’t have the luxury of time necessary to collect full drafts, comment on them in any significant way, and repeat again with the final drafts.  Fortunately, this is not the only way to create space for revision. Consider these alternatives, none of which are mutually exclusive:

  • Simply spend some time in class discussing revision, giving students examples of what you would consider a solid first draft and solid final draft and offering various techniques for getting from the former to the latter (there are many, many resources for revision advice to be found online–find one that feels like the best fit for your assignment and point students there, or ask the WAC Director for suggestions).  Have students submit their rough drafts along with the final versions, and be sure to comment on how well the final draft improves on the earlier version in your feedback to the students.
  • Conduct a peer review session, either in class or asynchronously through the cloud, in which students comment on each other’s drafts.  When they submit their final drafts, ask students to address if and how they responded to their peers’ comments.
  • Set aside all or part of a class session on the day drafts are due, and have students evaluate their own drafts: what turned out well, what aspects are still in process, and how will they proceed with that knowledge? Have them turn this into a written revision plan, to guide the process to the final draft.
  • In a class with multiple essay assignments, require students to choose one essay to revise and resubmit at the end of the term.  You can allow the revised essay to replace the grade of the earlier version, or make it a separate assignment grade (both approaches have their own pedagogical merits).

Each of these techniques helps to make revision an integral part of the assignment, not a side practice that we might encourage, but not explicitly require. They also create opportunities for you as an instructor to step in and guide students’ revision processes, helping them to see methods and opportunities to improve their writing that they don’t.  While this kind of feedback does require some time and attention on the instructor’s part, it can often allow for less feedback (and easier grading) on the final version of the assignment, as you’ll already have created a dialogue with the students around the assignment that you can simply bring to a close with a few explanatory notes and (ideally) a completed rubric.

The concrete ways that revision complicates the use of AI writing tools are fairly obvious. Essays produced by ChatGPT tend to be remarkably free from grammatical errors, but fairly vacuous; thus, they require little editing but a great deal of revision. And, since it’s generally harder to revise someone else’s work effectively than your own, students who produce their drafts with ChatGPT will often find that the revision process requires more work for less return.

Again, though, the real value of guided revision is that it encourages students to see their writing as a work in progress and to get a better sense of what they can accomplish with a fully developed writing process. The more pride and value they associate with their own writing, the less likely they’ll be to let AI do it for them.

Encouraging Reflection and Metacognition

The WAC Program has (to understate it mildly) promoted reflective and metacognitive writing as a teaching practice for some time, so it’s probably no surprise to see a section on them here. In the context of creating AI-resistant writing assignments, though, reflective and metacognitive writing take on additional layers of value, both because they help students to see the benefits of their own intellectual work and because they make it more difficult to conceal if they’ve allowed AI to do that work for them. 

Quickly stated, the distinction between the two modes is that reflective writing looks backwards (what have I learned and experienced?), while metacognitive writing looks forwards (how can these experiences inform my future actions and methods?).  These modes of thinking go together more often than not, and both are obviously essential for learning.  By challenging students to enact these processes–to actively consider, in writing, what they’ve learned and what they plan to do with that learning–we help them to connect the disparate pieces of their education, understand their own strengths and weaknesses more effectively, and simply develop a better understanding of themselves.

In practice, this is often a much simpler and smaller-scale operation than those principles would suggest.  Any formal writing assignment can be scaffolded fairly easily with one or more informal reflective and/or metacognitive tasks.  These tasks can (and generally should) be short and fairly simple both to write and to read.  They can happen at any point in the writing process–before, after, or even while the student is writing the assignment itself.  Consider a few common scaffolding assignments in this vein:

  • A week or more before a formal writing assignment is due, ask students to write a paragraph or two summarizing their topic and articulating why they chose it (ideally, in the context of the ideas above, explaining how their assignment will respond to ideas that they have encountered in class).  Then, ask them to assess the work they’ll need to do to complete the assignment–what sources will they engage and how, what questions will they have to address, what conclusions do they still need to form, etc.?
  • The weekend before the assignment is due, ask students to write a concrete plan for their entire writing process. How long do they think it will take to create an outline, write a draft, revise it, and proofread it? When and where will they do this work? What additional help, if any, will they seek out, from whom, and when?
  • As they turn in the assignment, have students compose a short note to you about their writing process and how it worked out for them. How did their ideas change between conception and completion? What steps were easy or difficult? Did any of that surprise them? Overall, how happy are they with the final product as they submit it to you? Comment on this note as part of your feedback to the student — based on the final product, what aspects of the student’s writing process served them well, and what practices might they want to change or adopt for the next assignment?
  • As they begin work on the next assignment, ask students to consider what they learned writing the last one. What ideas from that project will inform this one? What lessons about their own writing process did they learn, and how will that inform the way they write this assignment?

None of these techniques are particularly novel, but they are powerful, because they require students to consider not just the final product of the assignment, but the actual learning process that the assignment is designed to enact.  It seems obvious to us that we create writing assignments because we want students to learn from the process of writing them, but students tend to fixate on the product rather than the process.  This makes the actual work of writing seem obscure and magical, even when they do it themselves. They sit down at the computer, mull over the topic at hand, and gradually, sometimes painfully, the final product grinds itself into being.  When they’re asked to articulate these steps in writing, though, their process becomes visible to them, and the opportunity emerges for them to critically examine what they do and how they can do it better.

Their writing process also becomes more visible to you, which is useful both instructionally and in the context of making assignments AI-resistant.  In instructional terms, metacognitive writing allows you to respond to the way students work, not just the work itself.  You can see how their ideas evolve and, ideally, help to guide them.  You can see how students approach writing (from methodologies of research to basic time management) and offer encouragement or guidance.  In other words, you can make their learning an active part of the class.

This pushes against the encroachment of AI in a number of ways.  Most concretely, any of the metacognitive steps suggested above could be completed effectively in class, making it much harder (though perhaps not impossible) for students to covertly hand the work over to the robot.  Furthermore, if students know that they’ll have to explain their writing process in some detail, they might think twice about letting AI do the writing.  None of the tasks suggested above should be particularly challenging for students who actually completed the assignment as designed, but they’re considerably more challenging as exercises in creative fiction.  Pointing this out to students when you introduce the assignment can reduce the perceived benefits of letting AI do the heavy lifting.

On a less concrete but more profound level, though, this kind of scaffolding discourages AI-based cheating by placing value on the writing process rather than just the product.  By helping students see how the real intellectual labor called for by your assignment is meant to benefit them, you discourage the transactional mentality that sees assignments essentially as invoices for students to fill.  My experience is that the overwhelming majority of students are more willing to work than we tend to assume, as long as they believe that their efforts will be rewarded in some way, and metacognitive writing can help them see the benefits of their work in a useful and tangible form.

The same can really be said for any of the practices described above. At the end of the day, we cannot force students to feel invested in their educations, nor can we stop them from seeking out ways to make their academic work easier, legitimately or otherwise. However, if we start from the assumption that most students are invested in their education and truly want to learn, and we create assignments that allow them to see and experience that learning as it happens, then the perceived value of letting AI do their writing for them will diminish significantly. It will also allow us, as instructors, to approach AI writing tools from a position of strength by making it a teaching issue rather than a detection and enforcement issue.