Students Are Using ChatGPT to Cheat, But Professor Says AI-Written Assignments Can Be Easy to Spot

As artificial intelligence tools such as ChatGPT become increasingly common in education, concerns over students using them to complete assignments have grown. However, Johns Hopkins University professor Steve Hanke believes identifying AI-generated work is often easier than students expect.

Hanke, a professor of applied economics who has taught at Johns Hopkins University for nearly six decades, told Business Insider that he can distinguish between a student's own work and AI-generated writing by looking at how closely an assignment matches the student's usual abilities.

Sudden improvement can raise questions

According to Hanke, one of the clearest indicators is a sudden change in a student's writing quality.

A student who normally struggles with writing but submits an unusually polished, sophisticated or technically advanced assignment may immediately attract attention. Hanke said his long experience teaching students gives him a useful benchmark for assessing their normal writing and understanding.

He described himself as an “old fox” when it comes to identifying work that does not match a student's established abilities.

Knowing students helps teachers spot AI use

Hanke's approach does not depend solely on identifying characteristics of AI-generated text. Instead, he believes teachers' familiarity with their students can be an important factor.

After teaching students over a period of time, professors develop an understanding of how they write, reason and explain concepts. A submission that appears significantly beyond that student's demonstrated knowledge or writing ability can therefore raise questions.

Hanke does not consider AI itself the problem

Despite his concerns about academic misuse, Hanke said he does not view AI as inherently harmful to education.

Instead, he believes the widespread use of AI requires teachers to become more vigilant and pay greater attention to their students' individual abilities.

As AI tools become increasingly embedded in classrooms, the challenge for educators may therefore be less about detecting a particular technology and more about understanding how students normally think, write and learn.