Does the Unessay Help or Hinder Teachers on the Hunt for their White Whale? (Part 2)
- Dustin Rimmey
- 3 hours ago
- 5 min read
On Monday, we explored the concept of the unessay. In short, the unessay is an assignment concept forwarded by Daniel Paul O'Donnell in 2012. O'Donnell used unessays to question the overall role that formalized academic essays had played in education throughout the history of the system. Specifically, he argued that traditional essays are a "static and rule-bound monster" that forces students to comply rather than explore and showcase their intellectual passions.
The goal of Monday's post was to conceptually introduce the unessay and explore how teachers have been deploying them in the classroom to push students to complete tasks in a more creative fashion. Today's post explores whether or not the unessay is an assignment that can be relatively AI-resistant (or AI-resilient...I've heard people use both terms for the same thing). Is it possible for AI to complete the assignment in its entirety, or is the capacity of AI used for the assignment limited in scope or impact on the final product's creation?
Redesigning the Unessay in the Modern Classroom
This is incredibly tricky because AI can do almost anything. I'm not saying it does it well...but it can do anything. From writing podcast scripts (and recording them in some cases) to generative art and video to coding...how can we still measure student learning in these cases where AI can take on the entire creative process from tip to tail? The answer comes in shifting from the product to the process (I'm aware of my cliche again). We shift from the product to the process, the personal, and the physical.
When rethinking the unessay, these four suggestions for redesign allow you to create AI-resilient/resistant tasks for your students
Go Process-Heavy
If you only grade the final product of the unessay, you leave the door wide open for AI generation and nothing else, because an AI can create a digital zine, podcast, or collage in a matter of seconds. Instead of making the final product being 90% or 100% of the grade, make it 40%. The other 60% of the graded value comes from a documented journey.
The Proposal: Have students create a specific pitch of what they plan to make and why. Even if they can generate part of this, they are documenting the choices they plan to make before they begin. In some versions of the unessay, I've seen teachers require that the proposal be delivered to the teacher in a presentation or in a conversation. While AI may be able to do some of the work, students have to defend what is generated. This requires that students do more than prompt and turn it in; it requires some human-in-the-loop decision-making, which requires an early defense of their ideal choices.
The Annotated Bibliography: Have them submit a list of sources with quotes, notes, and more. While some AI agents have gotten better at sourcing, etc., they do not always cite sources legitimately; they can still hallucinate or fabricate sources, and it holds the student to answer questions about the research they have done for credit.
The "Making Of" Reflection: Have the students create a behind-the-scenes journal, short video, or daily log recording their journey. If they hit a roadblock, how did they overcome it? What creative decisions did they make? If they had to evolve from their original proposal, what changes did they make and why? While AI can fake a final product, it is difficult for AI to fake a messy, iterative description of a learning process.
Be Hyper-Local or Hyper-Specific
This one may seem antithetical to the unessay, since we ask students to explore broad connections through their own creative interests, but finding ways to be local or specific may hinder the effectiveness of using any AI/LLM. While AI may be successful in exploring global concepts, it is terrible at knowing what happened in my classroom last Tuesday.
You could design the unessay to explore explicit connections between your academic content and something localized, like:
A specific debate or discussion from an in-class seminar.
A local community issue
A synthesis of a specific guest lecture or discussion with your course readings.
When the prompting of an AI agent requires tying moments that aren't internet-accessible to the academic context, students are forced to do the cognitive synthesis themselves.
The "Live Defense," or an Interactive Component
Even the most elaborate, AI-assisted digital project falls apart if the student doesn't actually understand it.
Borrowing from the tradition of graduate thesis defenses, an AI-resilient un-essay includes a mandatory, unscripted "live" component. This could be a five-minute in-class Q&A about their board game's mechanics, a gallery walk where students must verbally explain their infographic to peers, or an oral reflection with the instructor. If the student used AI as a crutch rather than a tool, the live defense will immediately reveal the gaps in their actual comprehension. If you have students who are gun-shy in presenting to larger groups, have them record a video presentation to share with only you!
Lean into the Skid! Critique an AI Product!
This is one of the phrases that I never understood until I really had to deal with it. Here's a terrible AI-generated video I found on YouTube for those of you who rarely have to drive in heavy rain or snow.
Instead of chasing the white whale of AI-generated assignments, what if you required students to explicitly create an AI-unessay?
I know that some of you may freak out at what I just wrote....but stay with me!
Have the students create an artifact explicitly with AI, and then the students take on a role shift. They shift from being the "author" of the product to an "editor" or "auditor." They now must take the AI-generated product and extensively annotate it. They should point out hallucinations, shallow arguments, correct a bias, etc. Then, have the students shift again. They now must take all of their notes into mind and redesign the final product to create something that is both factually accurate and academically rigorous. This transforms the student from a passive consumer into an active, critical evaluator. Teaching your content + AI literacy in one loop, nicely done!
Parting Thoughts
The reality of education in the generative AI era is that we can no longer rely on the form of an assignment to guarantee its authenticity. Whether it’s a five-page analytical paper or a five-minute audio documentary, if the assessment focuses solely on the final product, an AI can likely generate it.
The un-essay remains a brilliant pedagogical tool, just as Daniel Paul O’Donnell envisioned it over a decade ago. It breathes life into tired syllabi, empowers students to leverage their unique strengths, and connects dense academic research to the real world in meaningful ways. But as a standalone defense against artificial intelligence, it falls short.
The path forward isn't about finding the perfect, unhackable assignment. It’s about shifting our focus from policing to pedagogy. By embracing the un-essay's creative freedom while grounding it in documented processes, hyper-local contexts, and live human interaction, we stop trying to out-tech the algorithm and start out-humaning it.
When we prioritize the messy, iterative, and deeply personal journey of learning over a perfectly polished final deliverable, we do more than just deter cheating. We create assignments that students actually want to do—and intrinsic motivation will always be the best AI-proofing tool we have.
In Friday's post, we'll look at the flipside of the argument and see if we are doomed to constantly hunt our whale, or if we can finally enjoy some smooth sailing.
Comments