“I’m very fond of the take-home essay, as there’s something irreplaceable about the experience of articulating a theory over the course of multiple weeks—doing background research, letting the ideas marinate in one’s subconsciousness, and chiseling away at the draft until every word is perfectly placed.”

That’s Tom Kaspers (University of Chicago), expressing a feeling many professors have about teaching in the early days of the AI era.

Yet unlike many who have either ignored their students’ AI use on take-home essays, or have eliminated that kind of assignment from their courses, Kaspers “set out to try to save this experience.”

In the guest post, below, a follow-up to a piece he wrote for the Boston Globe, he describes how he did it. As he notes, his method won’t work for all kinds of courses, but it may work for some.

[A “Cadavre Exquis” by Jacqueline Lamba, Yves Tanguy, and André Breton]

Writing Together: A Teaching Experiment
by Tom Kaspers

This past winter, I developed a new kind of essay assignment with my students at the University of Chicago. I was teaching a first-year philosophy class that doubled as an academic writing course; its primary purpose was to teach writing a philosophy essay. I wrote a piece for the Boston Globe on why I felt the need to change up the assignment and on how my students rose to the challenge and responded wonderfully to this little teaching experiment.

The reason I felt I couldn’t assign individual essays anymore was that I worried it was no longer an effective way to teach philosophical writing (or even academic writing more generally). We might sometimes forget that philosophical writing belongs to a very peculiar genre, upheld by a myriad of conventions, the bulk of which are unconceptualized. Teaching someone how to write a philosophy essay feels at times like teaching a foreign language. And for some students, you might as well be speaking Latin. The only way they’ll learn is through repeating cycles of writing and revising. But now they can skip that altogether by using tools like ChatGPT as a kind of Google Translate for their philosophical ideas: in go some half-baked philosophical thoughts, out comes a perfectly shaped essay. The only downside is that they’ll never learn to speak the language of philosophy.

Of course, that’s quite the loss, as the language of philosophy isn’t just some quaint provincial dialect; it is a vernacular that’s uniquely attuned to expressing abstract thought as clearly as possible. Moreover, doing philosophy is writing philosophy—they are often one and the same activity. Philosophy is all about achieving clarity of thought. So, the real philosophy happens when we take our half-baked philosophical ideas and turn them into defensible theories. Sadly, that’s exactly the task that students are outsourcing to their LLMs.

We can try to prevent this in a number of ways, from oral tests to handwritten exams. But I’m very fond of the take-home essay, as there’s something irreplaceable about the experience of articulating a theory over the course of multiple weeks—doing background research, letting the ideas marinate in one’s subconsciousness, and chiseling away at the draft until every word is perfectly placed. So, I set out to try to save this experience.

As I was thinking about alternative versions of the essay assignment, I realized that the best way for me to teach my students to write a philosophy essay was by simply showing them how it’s done. I decided to ask my students if they wanted to drop the individual essay assignments and instead write one big philosophy essay (~10,000 words) with me as their coauthor. I felt it was important to grant my students authority not just over whether to do this assignment at all but also over the exact shape of the assignment, as I really needed them to be on board with it, and as I believe more generally that students learn much better when they believe their learning experience is authentically theirs.

Since the appearance of the Boston Globe piece, I’ve received reactions from many educators at different levels (high school, college, professional schools), in different fields (e.g., film, economics, ecology, medicine, law), and from all over the world. Many wanted to know more about the nuts and bolts of the assignment. There were many details I had to omit to fit the narrative structure of the piece. I’m hoping to share some of these with you now.

I should start by mentioning that this assignment likely won’t work for all or really even most classrooms. As a teacher at the University of Chicago, I’m used to relatively small class sizes. I had 19 students, which I felt was already close to upper limit, as we’d want every student to meaningfully contribute without it devolving into chaos. I should also say that this was very much an experiment, so we made most of it up as we went along (including the grading).

We only had nine weeks to think up and write down an entire essay, which made for a tight schedule. First, I had the students brainstorm essay ideas in class, which they wrote down on a shared Google Doc. Then, we went through the ideas together and decided which were promising enough to be worked into an essay proposal. The students divided themselves up into groups and worked on these proposals. We discussed them in class and took a vote, after which I went to work to expand the winning proposal, adding a layout of a tentative argument structure, and a very lengthy list of possible sources. I had the students write their names beside these sources and prepare to discuss them in class. Additionally, they sent me brief emails describing how what they read might (or might not) be relevant to our essay.

The students decided for themselves on which parts of the essay the wanted to work. Most of them jumped between various parts. In this early drafting phase, they sent their thoughts directly to me, and we decided together how they could best contribute to the essay. I think that having a detailed argument structure with clearly delineated parts is absolutely key to making this phase work. At the same time, though, I did appreciate that none of my students stuck with only one part of the essay; they all ended up thinking about the essay holistically and trying to grasp the bigger picture. I created a very rough first draft from the contributions of my students. They then split into groups and started working on different sections of the draft. Their contributions still went through me; they’d send me a new draft of a section, I’d tell them it’s no good, they’d revise it, I’d tell them it’s still not quite there, etc. In the final phase of the assignment, I allowed the students to directly edit and write into the essay.

Pretty much all of our class time was devoted to discussing the essay. I absolutely loved this, as every student was there and ready to participate. In fairness, they kind of had to be, as it was usually in class that we decided on how to proceed with the essay—what direction to take it in, which parts to revise, which authors to discuss, etc. But a lot happened out of class too, as students were constantly emailing me with their ideas.

This probably had something to do with how we decided to grade them. They wanted to be graded on their individual contributions. I split the grading into three components: brainstorming, background reading, and writing. But I also told them that they could contribute in many different ways, and that they could compensate for one task by doing another one, e.g., someone who wasn’t as influential in the brainstorming phase could make up for it by doing a lot of writing. It was relatively easy to track both the quantity and quality of their contributions, as most of it had gone directly through me. For the work they did in groups, they self-reported on the division of labor. Only in the last phase did it get more complicated, as I had to go through the history of our shared Google Doc to see who did what. But at that stage some students had already secured As, so I only had to keep track of a subset of my students.

I know this assignment doesn’t totally destroy the potential for illegitimately using artificial intelligence. The idea was just to put some of the friction back into the learning experience—the friction that’s lost when ChatGPT takes over the writing process. The fact that we were in a continuous conversation about the essay—both in and outside the classroom—meant that students constantly had to defend their writing and their ideas. They had to pay attention in class to understand what they needed to do to contribute. And we ended up writing many passages truly together, in the classroom. This meant I still got to show them how to write in the language of philosophy. Therefore, even if the friction didn’t deter them from using AI, it still served its purpose.

Note: Here is an ungated version of the Boston Globe piece.