Whenever I share my writing, people often accuse me of using AI. The accusation rarely targets my ideas or argument. Instead, it attacks my writing style, specifically my long sentences and em dashes. To put it simply, I write like AI because I was trained with the same material.

I was homeschooled with a classical education program, which meant I didn’t read the same material as my public-schooled peers. In high school, I read old novels, essays, and speeches, and was taught to notice how a sentence was built: where it paused and where a writer let a thought run long. These habits became how I put words on a page, especially in my prose. I also finished grade school before AI tools became part of everyday life, so my writing style was already formed before anyone worried about AI taking our jobs.

Ironically, the same books that shaped my language also shaped AI. Large language models such as ChatGPT, Claude, and Copilot are trained on enormous amounts of text: books, articles, and writing gathered from across the public web, much of it taken without the consent of the people who wrote it. Much of the polished prose in those datasets use em dashes freely because skilled English-language writers have always used them. If you read the work of essayists, novelists, and journalists of the last two centuries, you will find the em dash everywhere.

AI predicts the next word from statistical patterns, while a human writer paces a sentence with intent. When a writer uses an em dash, it is because they want to write using an interruption. When a model produces an em dash, it reproduces a pattern it saw in human work. A model is not a person. The model learned the dash from human writers; and humans did not learn it from the model.

A common assumption is that short, plain sentences are honest, and anything more elaborate must be AI. Plain writing is valuable, and I use it when my work calls for it. Complexity isn't wrong, and writers shouldn't be punished for it. I should not have to dumb down my writing for audiences, especially in writing spaces. If we treat proper syntax as suspicious, we don't make writing more human; we teach young writers that ambition on the page is taboo.

There is also a practical problem with accusing people of using AI based on how their writing looks. Nobody can reliably tell—and that includes the people who are most confident they can. Style is a poor fingerprint because so many human writers share the habits that models learned from them. A reader who spots a semicolon and declares the piece machine-made isn't detecting anything; they are guessing, and the person on the receiving end pays for it. As a student, an accusation like this can mean a lost grade, a lost scholarship, or a damaged reputation, all over something as ordinary as punctuation. Doubt is easy to cast and very hard to disprove. If I could hand over a certificate proving that I wrote every word myself, I would, but since I can't, all I can do is say plainly that I did. Anyone is more than welcome to ask for my Google Docs history.

If you want to know whether someone wrote their own work, ask about it. Ask what they were trying to say, why they structured the piece that way, or what they cut. A person who made it can talk about it as long as you like, and that conversation will tell you far more than an arbitrary count of em dashes ever could.

A reader who sees complex language and assumes AI usage tells us more about their reading habits than about ours. If most of what someone reads is flat and fast, anything with rhythm will feel strange to them, and they can easily mistake that strangeness for artificiality. That assumption belongs to the reader, not the writer. I am going to keep writing the way I always have. “Strange” punctuation is not always an indicator of AI use. It is a choice I am allowed to make, and I'm going to keep making it.