Most Grad School Applicants Use AI Essays — And It Backfires

A new study looked at nearly 7,500 applications sent to a large public policy master's program in the United States between 2020 and 2025. The goal was simple: figure out how many people are using AI (chatbots like ChatGPT that write text for you) on their admissions essays, and whether it helps or hurts them. The answer turns out to be both.

The timing gave researchers a natural experiment. ChatGPT launched in November 2022, creating a clear before-and-after line. After that date, essay writing quality went up measurably. But here's the twist: in the 2025 admissions cycle, the majority of applicants submitted at least one essay that was mostly AI-generated — even though the program explicitly told them not to. And those applicants were admitted less often than similar people who wrote their own essays.

To find out why, the researchers ran a separate experiment with real admissions officers. Officers could often tell when an essay was AI-written, and when they suspected it, they scored it lower. So the penalty isn't just about the words themselves — it's about the suspicion they trigger.

For anyone applying to school, the takeaway is uncomfortable. AI can polish your writing, but it may also stamp your essay with a signature that readers recognize and quietly punish. The paper also raises a bigger question: admissions rules and review processes were built for a world without AI. If nearly everyone is using it, schools may need new policies rather than bans that most applicants ignore.