Steve Hanke AI cheating is, in his own telling, a problem he finds relatively straightforward to manage: the Johns Hopkins University professor of applied economics says he has ‘absolutely no problem’ identifying when a student has submitted machine-generated work, and he puts his confidence down to decades of accumulated classroom instinct.
The ‘old fox’ and the chatbot
Hanke, who has been teaching at Johns Hopkins for nearly 60 years, described himself to Business Insider as an ‘old fox’ who finds it ‘very easy’ to tell a student’s own writing apart from a chatbot’s output. The diagnosis he offers is blunt: today’s students, even at elite universities, have ‘poor writing skills,’ and that baseline makes AI-generated prose stand out rather than blend in.
‘It’s blatant when a student submission has been generated by AI,’ he said. Beyond prose style, Hanke also draws on his sense of where each student stands in their understanding of economics. When a piece of work fails to match that level, he said, it becomes ‘pretty easy to spot.’
He was careful not to dismiss the challenge entirely. ‘The only problem that AI poses is that it requires the old fox to be on guard, more than would have been the case sans AI,’ he said. The concession is modest but telling: the arms race between students and instructors is real, even if Hanke rates his own position in it highly.
Hanke served on President Reagan’s Council of Economic Advisers. That career arc, spanning government advisory roles and nearly six decades of university teaching, underpins the confidence with which he approaches the question of Steve Hanke AI cheating detection. A professor who has read tens of thousands of student papers across multiple generations has a calibrated sense of what undergraduate economics writing actually looks like.
A view shared by admissions specialists
Hanke is not alone in his assessment. Drusilla Blackman, a former dean of admissions at Harvard and Columbia, echoed his position in a separate interview with Business Insider. Blackman said it is almost always ‘evident’ to an educator when a student has used AI, because what they submit does not match their usual standard of work, writing, or critical thinking.
‘If they write something that is not proportional to that, it is detectible,’ said Blackman, who is the founder of Deans of Admissions, a group that advises families on university entry in the United States and Britain. The argument she makes is essentially the same as Hanke’s: educators build a longitudinal picture of each student, and a sudden uplift in quality or a shift in register registers immediately as anomalous.
Both perspectives rest on a form of detection that no AI tool can straightforwardly defeat. Software-based AI detectors have attracted considerable scepticism for both false positives and false negatives. Human pattern recognition, anchored in prior knowledge of a specific student, is a different instrument altogether, and one that scales poorly as a general institutional solution but works well for individual instructors who know their students well.
How some educators are responding beyond detection
Not every instructor shares Hanke’s equanimity about simply identifying Steve Hanke AI cheating-style submissions after the fact. Business Insider previously reported that several teachers were using AI defensively, designing assignments intended to be resistant to chatbot completion. Others have reverted to handwritten work to remove the tools from the equation before a submission is even made.
Those two responses point to a structural tension in how institutions are handling generative AI: whether to build better detection, or to redesign assessment so that detection becomes less necessary. Hanke’s approach, grounded in close knowledge of individual students accumulated over decades, sits closer to the former, and depends on a depth of familiarity with learners that not every teaching context can replicate.