Johns Hopkins Economist Says AI Cheating Is Easy to Catch — Because Students Write So Poorly

A Johns Hopkins economics professor with nearly six decades in the classroom says AI-assisted cheating poses little challenge to experienced educators — and the reason is somewhat ironic: today's students write so badly that polished, AI-generated prose stands out immediately. Steve Hanke, a professor of applied economics who served on President Reagan's Council of Economic Advisers, says it's 'blatant' when a submission has been machine-generated, precisely because it doesn't match what students typically produce.

Hanke describes himself as an 'old fox' who has developed a keen sense of his students' economic knowledge and writing ability over decades. That baseline makes it straightforward to flag work that doesn't align with a given student's demonstrated skill level — even at an elite institution like Johns Hopkins. His message isn't alarm, but vigilance: AI requires experienced professors to stay sharper than they might have otherwise.

Former Harvard and Columbia admissions dean Drusilla Blackman, now founder of the college advisory group Deans of Admissions, independently reached the same conclusion. She argues that when a student submits work that is disproportionately sophisticated compared to their established record of writing and critical thinking, it's almost always detectable. The mismatch, she says, is the tell.

Beyond human detection, educators are also adapting structurally. Multiple teachers have begun designing AI-resistant assignments — tasks that require real-time reasoning, personal experience, or in-class demonstration — and some have reverted entirely to handwritten work as a low-tech countermeasure to the chatbot wave.