Last summer, I felt crushed by AI.

I taught the first half of the U.S. history survey—100 students strong—fully online for summer school. I reused a series of short essay assignments that I’d crafted over the last decade. I liked the assignments. They were engaging and thoughtful. They asked students to analyze, synthesize, interpret sources, and think like a historian.

The first few papers I read were solid. They were tightly organized, answered the question, and offered an interpretation of the assigned reading. The tone was a little flat, but probably, I thought, that’s just the way these students write.

Then, the next five papers were the same. They had a five-paragraph structure, a relatively thorough discussion, and that detached, bloodless tone.

Then, the next dozen papers were similar. I then noticed the papers had no errors in spelling, grammar, or syntax. They were technically flawless. The essays were also 20 per cent longer than usual. Had students suddenly gotten a lot better at writing?

I started running my questions through ChatGPT and had the answer—and a sinking feeling in my gut. AI had gotten better than I thought possible, faster than I’d thought possible. And students had gotten better at prompting it.

What could I do? I’m too young to retire and too old to start a new career. After grieving my loss, I decided to do something radical. I threw out all my old writing prompts—laid waste by AI technology—and started from scratch. I began by asking myself what I wanted writing assignments to accomplish.

My simple answer: I want students to think, and writing is a powerful tool that makes thinking possible.

That answer felt too vague, so I tweaked it a bit. I wanted students to think about the tension between the familiarity and the strangeness of history. It’s natural, I tell students, to read about some historical person or event and think “These guys are so weird—everything was different back then!” It’s also natural to think “These guys are just like people today—nothing really changes!” Both reactions can be true. Wrestling with that tension is a task worth thinking and writing about.

Fortunately, this goal for writing assignments called for the kind of answer AI doesn’t excel at yet, namely, writing from a personal perspective. As a result, I’ve recast all my writing assignments to ask students for a personal perspective on some historical person or event, supported, of course, by evidence from assigned course materials.

Here are some examples of the assignments I’ve created.

For the colonial period, I assigned short selections from David Hackett Fisher’s Albion’s Seed: Four British Folkways in America (Oxford, UK, 1989), and asked about marriage practices. The question: Drawing from your reading of David Hackett Fischer’s discussion of “marriage ways” in Puritan New England and in Anglican Virginia, which region’s approach do you find most appealing? Explain why.

Students talk about their own beliefs about love and marriage and how their beliefs are reflected (or not) in colonial cultures.

For the American Revolution, I use Benjamin Irvin’s New England Quarterly article “Tar, Feathers, and the Enemies of American Liberties, 1768–1776” (Vol. 76, June 2003, 197–238), and ask about the morality of tar and feathering. The question: If you were a person living in a city like Boston, Philadelphia, or New York, would you have participated in a tar-and-feathering of a British official or loyalist civilian? Drawing from the article “Tar, Feathers, and the Enemies of American Liberties” explain why or why not.

Students have to decide what kinds of protest are acceptable and under what circumstances they would act.

“The Bostonians Paying the Excise-Man” — a 1774 British print by Philip Dawe that depicts the tarring and feathering of Boston Commissioner of Customs John Malcolm (Wikimedia Commons).

This summer I’ve gotten more creative, and I’ve started writing fictional scenarios that reflect real dilemmas people faced.

For the early republic, I took a chapter from Catherine Allgor’s biography of Dolley Madison (Dolley Madison: The Problem of National Unity, New York, 2012) on the Merry Affair, an 1803 diplomatic controversy triggered by President Jefferson’s democratic style clashing with the British ambassador’s expectations of formality for himself and his wife. The question: Imagine you are a friend of Thomas Jefferson, such as James or Dolley Madison. How would you advise him to handle the Merrys? For example, should he hold firm to his more casual social style? Or adopt a more European style to placate the Merrys?

I like this question because it uses an event students have never heard of and makes it meaningful. They must consider both international politics, which are likely remote for students, and interpersonal relationships, which they navigate all the time.

Gilbert Stuart’s 1804 painting of Dolley Payne Madison (Wikimedia Commons).

For my current summer class, I’ve created an assignment where I ask about dueling in the Old South using a chapter from Kenneth Greenberg’s book Honor and Slavery (Princeton, NJ, 1997). I try to place students inside the southern culture Greenberg describes. The question: It is a warm summer evening in 1835, and you are sitting with your fiancé on the porch of your family plantation outside Tallahassee. He shares some difficult news. He has been challenged to a duel. It all started when he accidentally bumped a man in a crowded theater, and he failed to immediately apologize. The situation quickly escalated. Your fiancé doesn’t know how to respond. His fear of humiliation should he refuse to duel is matched by his fear of losing you should he be killed. What would you advise him to do?

I haven’t graded the students’ submissions yet. We’ll see how they turn out, but overall, I’m satisfied with my reformed assignments. Students are engaged. Their writing shows evidence of thinking. Students, it seems, are doing more than copying and pasting from ChatGPT.

Of course, I don’t truly know how many students are completing assignments themselves and how many are using some form of AI. I suspect there are quite a few papers where students used AI somewhere in their process but also engaged in their own thinking, too. Right now, I don’t know how to tell where student writing ends and AI begins.

Sadly, the scenario prompts that are so much fun for me to write appear especially vulnerable to AI use. AI can easily create fictional characters, like a woman and her fiancé sitting on a porch in Tallahassee, and write a scene, complete with plausible dialog for their conversation. I didn’t appreciate that move in advance. I’ve decided I’ll have to ban in-character responses next semester.

I have no illusions that asking for a first-person perspective solves the problem of students using AI to replace their own thinking in their written work. Students can prompt AI to write as if it is a college student, after all. However, I can see a difference between papers that simply tag a few “I” statements onto a conventional five-paragraph paper and essays that more fully incorporate a student’s unique life experiences. The student who talks about growing up in the Bahamas as part of why he would fight for the Union in the Civil War is different from a student who turns in a paper that describes the three major motivations of Civil War soldiers before wrapping things up with a quick “I’d definitely sign up to fight against slavery,”

Looking back over the last year, I’ve accomplished my goal of making marginal improvements. I can’t help students who are determined to take shortcuts. Instead, I’ve aimed to persuade students who want to learn something, if the task is clear enough, engaging enough, and well enough supported by faculty that the assignment appears manageable.

In our current AI world, marginal improvements, small steps to a better world, are good enough for me.