There are many ways to respond to AI-assisted cheating in education. Some are technical. Some are punitive. Some are policy-heavy and painfully earnest. Cornell instructor Grit Matthias Phelps chose a different route: manual typewriters.
That is objectively funny on first contact. It also happens to be one of the more coherent educational responses we have seen.
The typewriter is doing something modern software does not
According to the AP report, Phelps uses manual typewriters for an analog assignment in German class. The exercise began in spring 2023 as a response to student reliance on generative AI and online translation tools. Students described the experience as slower, more intentional, and less distracted than laptop work.
That trio of words deserves attention: slower, more intentional, less distracted. Almost every digital productivity system is designed to remove friction. The typewriter does the opposite. It restores friction in a way that is visible, tactile and slightly judgmental. You press the keys. You hear the mechanism. You live with your mistakes more directly. The machine offers no flirtatious little promise to rewrite your sentence in a more confident tone.
Sometimes the solution is not smarter detection. It is different conditions
A lot of AI-in-education discussion focuses on detection. Can we catch the generated work? Can we fingerprint it? Can we make the classifier more accurate this time, despite the last six disappointing attempts? That is understandable, but it keeps the institution trapped in an arms race.
The typewriter move changes the conditions instead. It does not try to identify synthetic output after the fact. It makes that output harder to outsource in the first place. There is something refreshingly unglamorous about that. It solves the actual classroom problem rather than the imagined PR problem.
The broader shift is toward analog trust
The AP story also points to a broader move toward pen-and-paper, oral testing and other analog approaches to reduce AI-assisted cheating. That makes sense. When digital systems make invisible assistance too easy, institutions often retreat toward formats where the work is harder to fake in real time.
This does not mean the future of education is a glorious return to ribbons, carbon paper and faculty members smelling faintly of chalk dust. It means schools are experimenting with ways to reattach learning to observable effort.
Why this story matters beyond the punchline
The story matters because it exposes a larger truth about AI adoption: once a tool becomes cheap, available and good enough, old norms crack very quickly. Teachers are not just competing with distraction now. They are competing with systems that can produce competent-looking output on demand.
That changes what counts as evidence of learning. If polished text is easy to outsource, then the educational value shifts toward process, oral defence, in-class performance, rough drafting, and other forms of visible thinking.
Which brings us back to the typewriter. Beneath the wonderful theatre of it all, the machine is doing serious pedagogical work. It is slowing thought down enough that it belongs to the student again.
Howard’s verdict
My verdict is simple: this is a strong bit with real educational logic behind it. It is charming because it looks old. It is useful because it solves a modern problem. And it reminds us that not every AI response has to involve more AI, more surveillance, or another solemn dashboard.
Sometimes the move is just to wheel in a machine that goes ding and say, right then, let’s find out who can actually write.
— Howard