Around the third stack of essays in October, she became aware of it. She had taught The Great Gatsby for nineteen years, and there was something strange about the papers that she couldn’t quite put her finger on. The sentences were well-structured. The changes were successful. Suddenly, half of her sophomores had grown fond of the semicolon, even the ones who typically submitted single-paragraph drafts written in the hallway.
Here, her name is irrelevant, and she requested that her district—somewhere in the suburbs outside of Columbus—be kept anonymous. At a public high school with roughly 1,400 pupils, she teaches English to tenth graders. The word that came to mind when she talked about that October afternoon, sitting at her kitchen table with a glass of wine and a red pen, was uncanny. Not incorrect. Simply uncanny. It’s like witnessing someone you’ve never met rearrange a familiar room.
It’s easy to argue that educators should have anticipated this, and the majority of them did in theory. After ChatGPT’s launch in late 2022, a nervous slideshow about academic integrity appeared at every American staff meeting by the following spring. However, being aware of something’s existence is not the same as seeing it take effect in your classroom. She acknowledged that she had mostly assumed her students wouldn’t bother during the first year. They were sixteen years old. Their chargers were left behind. They couldn’t possibly be operating a sophisticated essay-laundering scheme.
They were. Or at least half of them were. She began keeping a tally after that October stack. By November, she calculated that ChatGPT or one of its cousins had written between 40 and 60 percent of the essays that had been submitted. She described the detection tools her district had purchased as “almost worse than nothing” because they flagged the meticulous work of her honors students and eliminated the blatant fakes. Researchers at Iowa State discovered that even confident judges make mistakes when identifying AI authorship, with teachers correctly identifying it only about 70% of the time. That was what she thought. The self-assured judge had been her.
In the end, it wasn’t the polish that exposed them. It was the lack of it. After writing about her grandmother’s hands for the entire semester, a student abruptly submitted a Gatsby essay that discussed “the broader implications of the American Dream as a sociocultural construct.” Three pages of fluid, anodyne prose with absolutely no opinion were written by a boy who was unable to remain still long enough to read a chapter. The essays were accurate. There was nobody in them.
She began bringing the writing back into the classroom, something that many educators are now quietly doing. drafts written by hand. essays written in class. conferences where students are required to explain their ideas, sentence by sentence. A few children were clearly at ease. Others appeared stunned. One girl, a strong student, broke down in tears and expressed uncertainty about her ability to write without it. She claimed that she was devastated for a few days by that conversation.
Since the incentives haven’t changed and technology continues to advance, there isn’t a satisfying conclusion to this tale. According to Ethan Mollick, a professor at Wharton who frequently writes about AI in education, students who rely on ChatGPT for their homework report feeling as though they are learning more while actually performing worse on tests. Teachers are aware of this. Most of the students are also aware. Now that the simple version is just a click away, no one seems to understand the purpose of high school English.
She continues to teach. She continues to use the red pen. She simply has less faith in it.
