How to Stop AI Cheating

4 pseudolus 2 5/29/2025, 10:47:27 AM honest-broker.com ↗

Comments (2)

TimorousBestie · 19h ago
His solutions:

1. Handwrite everything — does nothing to prevent AI use. Students already copy from ChatGPT by hand.

2. Oral exams & tutorials - doesn’t scale. The argument is universities could drop administrators in favor of more professors but more professors => more facilities => more capital expenditure. MIT can afford a student:prof of 3:1, but e.g. Ohio State can’t magically hire 3x more professors to go from 16:1 to 4:1. They’re starved for land as it is. Nor can they magically cut 75% of their student body and still afford the land they’ve got.

If they skimp on the facilities, then those new faculty are basically adjuncts with little prestige or authority. The main ingredient of the Oxford solution is prestige and a sense of authority and autonomy.

3. Handwritten/oral exams, pass fail

Oral exams are a nightmare to implement fairly. Oxford can probably afford to select for charisma and wealth signals over knowledge, but can Texas A&M? Oxford also has the prestige to brush off most accusations of improprieties or inequities in subjective grading; most other universities do not have that and cannot acquire it easily.

I do not believe for a second that the exam proctors of Oxford are incorruptible and cannot be bought at any price.

4. “Tough” grading or else the credential is devalued

I more or less agree on this point, though I will point out that American universities that have resisted grade inflation (Purdue, for example) do not seem to benefit from it in prestige. MIT has a 95% graduation rate, but few people argue Purdue is a better engineering school for failing about half its freshman class.

5. Only face-to-face comms

This point is just absurd. Oxford professors use email just like everyone else; I’ve emailed colleagues who work there.

This makes sense for an aristocratic finishing school but hardly generalizes to a STEM-focused school churning out middle-class professionals.

rvz · 20h ago
It cannot be stopped.

Going back to basics like what Oxford has always done is somewhat part of the solution, but does not solve the incoming age of the AR glasses.

The cat and mouse game in stopping AI cheating will only going to get worse for those who are digitizing their teaching.

Good luck.