“Robot will ruthlessly assess robot, while sentient minds gratefully retire from the scene.” (Erin Clark/The Boston Globe/Getty)
Imagine you discover a magical ring that bestows invisibility on its wearer. What do you do next? Many an introductory ethics course has started with the Ring of Gyges from Plato’s Republic: the tale of a shepherd who found such a ring, then “seduced the king’s wife, and … set upon the king, and slew him and possessed his kingdom”. As freshmen discuss the old thought experiment this month, some will harbor smaller, more doable ambitions: to fake their first essay convincingly, and perhaps every piece of work thereafter.
To obtain the prize that helped him lose his scruples, Plato’s shepherd had to descend into a chasm, enter a hollow bronze horse through a small door, then remove the ring from the finger of a corpse. Students today just have to ask Claude. AI agents can create a literature review from scratch; fabricate data sets; write computer code or solve equations. They can even check exam answers in real time. Lecturers are apparently powerless to detect the malfeasance. Those students who find the inner strength to avoid such shortcuts are watching peers sail ahead of them in the rankings, effort-free.
This makes cheating a particular problem in aggressively competitive elite universities. Writing in The Atlantic, behavioral scientist John Paul Rollert reports that a third of Harvard students now admit to using AI illicitly. But it is also a growing problem in the academy as a whole. An estimated 22% of UK-based students used it to cheat in 2023/24, and the numbers have surely increased since then. Rollert worries that we are creating a generation with “a knack for cutting corners, a disdain for community rules, and an assumption that a spirit of integrity… is the reserve of suckers and wimps”. Though there is some truth in this, it all sounds a bit too Patrick Bateman for the average situation. Most young cheats are just following the crowd, and aren’t gaining a knack for anything at all.
The classic trickster archetype has always been glamorously countercultural. Plato’s shepherd seems ingenious and daring, despite his murderous desires. But AI doesn’t even give us that small consolation. It spawns impressive-looking symbols and sentences at the click of a mouse, and the user can’t take any credit. He doesn’t even need to understand what he is passing off as his own.
Indeed, rather than exemplifying bracing tough-mindedness, I suspect that the typical student cheat feels swamped and powerless in the face of incomprehensible demands from teachers, and doesn’t know what else to do. As British universities continue to hoover up young people with lower A-level grades in order to pocket their fees, I would bet that some baffled young souls are moving from using AI as a search engine (harmless) to actively creating content (bad) without even properly noticing the switch.
Meanwhile, more deliberate types seem to be reasoning: if everyone else is using it, why shouldn’t I? And by the way: “everybody else” includes lecturers now. At King’s College London, using AI to produce student feedback is officially sanctioned. And at Cambridge, the main objection to using it for marking is that the tech is “not yet good enough”, rather than because human judgment and communication should be sacred. In this lax context, there are bound to be academics, faced with large numbers of scripts to process in a short time, who outsource the whole thing. Robot will ruthlessly assess robot, while sentient minds gratefully retire from the scene.
When everyone is cheating a system, the system collapses; or so another philosopher assumed, writing centuries after Plato. You should not lie or make false promises, Immanuel Kant told his students at the University of Königsberg, because if everyone did so, it would hollow out the point of truth-telling and promise-keeping: nobody would be believed or trusted about anything. Kant thought that moral principles like this could be extracted from the demands of rationality alone, though it was never very clear what was so irrational about a small number of deceivers free-riding off the basic honesty of the masses for personal gain. It might seem that student cheating presents a more compelling case than breaking promises. As it becomes close to a norm, it appears to edge the whole academic sector towards irrational systemic collapse.
Minus the Kantian trappings, this seems to be the most common form of objection to normalizing malfeasance. The main point of a modern university is to produce particularly knowledgeable, insightful people. When that sort of person is no longer being made, because robots are doing the hard yards instead, universities become redundant. For managers and teachers to tolerate AI-enhanced deception is therefore a catastrophically self-sabotaging act.
But reasonable as this objection initially sounds, I’m starting to feel a bit skeptical about the implied mission statement. If fostering cognitive excellence and producing clever people really was the main point of universities, wouldn’t the sector at least occasionally act like that was true? Instead, everywhere you look there are derogations from the presumed cause. From falling admission standards to rampant grade inflation, the impression is of a system focused on the superficial appearance of high standards, but unbothered about actually producing them. Rival concerns keep getting in the way.
Consider assessment itself. In-person exams are now mostly a thing of the past, having been substituted with online tasks — which of course make AI cheating much easier. The removal is said to benefit minorities, anxious people who hate exams, and those who have to fit university work round a job or health concerns. Few seem interested in the intellectual wins or losses per se. What skills and ways of thinking go unjudged and neglected once assessment is released from the old limits of time and space? Were they ever worth having, or not?
Or take the tyranny of specifying “learning outcomes” for every undergraduate module, and having to robotically match assessment and feedback to those outcomes. Originally mandated in the Nineties, this was always a terrible fit for the humanities, forcing teachers to treat free-ranging disciplines like English Literature or Philosophy as if their main contributions involved a core set of memorizable axioms or scientific techniques. Learning outcomes were not imposed because they produced more able students and brilliant thoughts, but for the purposes of “transparency” and “standardization” across the sector. In practice, they created mountains of bureaucracy, and made teaching a lot more dull.
Generally, it seems that wherever some other value — accessibility, cost-cutting, fairness, or whatever — comes into conflict with the goal of encouraging unusually bright people to have stonkingly good ideas, the latter is abandoned. And if that’s right, then the claim that an epidemic of cheating threatens the rationale of the academy doesn’t fly. If the main point these days is simply to give as many young people as possible a degree qualification, then perhaps it doesn’t matter exactly how they came by one. Or if the aim is “quality assurance”, so that all degrees in a given subject must be of the same standard, then why not simply pay for Claude for all?
To this, it will perhaps be objected that there is a much more obvious role for universities that is threatened by unbridled cheating: namely, readying young people for the jobs market. But even here, as AI wrings the life out of formerly demanding career paths, it is starting to matter less whether entrants can memorize complex material, think critically, or make perceptive inferences. If this is the best defense of honest toil that academic institutions can muster, they look vulnerable indeed.
Hegel famously complained of Kant’s ethics that abstract rationality cannot, in fact, produce a satisfying morality; it can only comment on means, not ends. There is nothing irrational about encouraging widespread lying and promise-breaking, if you don’t care about the shoring up of social trust. Equally, there is nothing self-defeating about turning a blind eye to cheating, assuming that the real goals of the institution are unaffected either way. To see off the incoming invisible academic crime wave, we will have to learn to talk without embarrassment about the intrinsic value of human erudition, and how universities, uniquely, can produce it; or else watch helplessly as the kingdom is conquered, not by shepherds but by the bots.
