Who Really Owns the Machine That Writes Your Essay?

So, a philosophy professor I heard about recently ran a plagiarism check on a student paper and found something odd. The essay was clean, with no matching sources, no red flags and also cumberground. It's like a fluent arrangement of nothing. It had clearly been generated by an AI. Of course, the professor failed the student for cheating. But, when a colleague pointed out that the same professor had used an AI tool to draft that semester's lecture slides, the conversation in the department got very quiet.

Here is the striking fact: at Middlebury College, a recent survey found that over 80% of students now use generative AI for coursework. Sounds like a faster adoption curve than almost any consumer technology in history, reached in under two years. Across the UK, a HEPI survey found AI use for assessments jumped from 53% of students to 88% in a single year. Nearly 7,000 UK university students were formally caught using AI to cheat in 2023–24 alone — triple the year before.

So, as a Student myself, I have a question: when a tool built from the collected knowledge of humanity gets enclosed by a handful of companies and rented back to the very students whose future labor it's supposed to prepare — who, exactly, is doing the cheating?

Core Problem

Universities are punishing individual students for using a form of knowledge that was never individual to begin with. Academic integrity policy assumes a lone mind producing original thought in isolation, and treats any outside cognitive assistance as theft. But a large language model (LLM) isn't really "outside" knowledge in that sense. As a compression of millions of books, papers, forum posts, and lines of code, most of it is produced collectively and anonymously, by people who will never see a dollar or a citation for it. When a student asks an AI to help explain a concept or draft a paragraph, they're drawing on what Karl Marx, in an 1858 manuscript, called the "general intellect" — the accumulated social knowledge of an entire civilization, increasingly folded into machines rather than held in any single head. Treating such as a private-property violation misunderstands what's actually happening.

Evidence

Look at where that "general intellect" actually comes from, and who profits from it. Training a model like ChatGPT required scraping and processing an ocean of human writing — most of it produced by people paid nothing for it. And, the dirty work of making these tools usable was outsourced to some of the lowest-paid labor in the world. A 2023 TIME investigation found that OpenAI paid Kenyan data-labelers, hired through an outsourcing firm, between $1.32 and $2 per hour to sort through the internet's worst content and make ChatGPT safe for classrooms — all the while OpenAI itself was paying that firm up to $12.50 an hour per worker. The knowledge, and the labor, were collective and cheap. The product is now sold back to students at $20 a month.

Universities have poured resources into catching students rather than reckoning with any of this. A University of Reading test found that 94% of AI-written exam submissions went completely undetected by human markers — meaning the detection arms race universities are fighting is one they are currently losing badly, at real cost to students falsely flagged and faculty time burned chasing ghosts.

Why It Happens

Not a confusion when it's a category error built into the system. Higher education still runs on a craft-labor model of knowledge: skill lives inside one person, and a degree certifies that this person, individually, can produce it on demand. That model made sense when knowledge production genuinely was artisanal, but it stopped matching reality once knowledge became something manufactured socially and stored in machines — first in libraries and search engines, and now in models trained on nearly everything humans have written.

There's also a less flattering incentive at work. A university degree is, among other things, a signal sold to employers about the reliability of a graduate's individual output. The cleanliness of that signal is threatened by AI. AI isn't necessarily threatening the learning underneath it. Much of the panic about "cheating" is really panic about whether the credential still means what it used to mean, which is a real institutional problem. It is not, however, the same problem as whether a student who used AI actually learned anything — and treating them identically is where the policy breaks down.

The Path Forward

Universities should stop trying to re-fence knowledge that has already become social, and start doing two things instead. First: redesign assessment around judgment, not unaided production. Evaluate how well a student can direct, question, and critically improve on AI output — the skill that will actually matter in every field they enter — instead of pretending the ideal student works in a sealed room with no tools.

Second, and more importantly: fight the enclosure, not just the symptom. Universities collectively spend abliguritious sums on AI subscriptions and detection software sold by the same handful of firms that built the walls in the first place. That money and institutional weight could instead go toward open, transparently trained, university- or publicly-governed AI models — treating the general intellect as the commons it actually is, rather than paying rent on it twice: once to train it, once to use it.

Closing

The philosophy professor grading that hollow essay wasn't wrong to be uneasy. Something genuinely is being lost when a student outsources thinking entirely to a machine. But the fix isn't nostalgia for a version of "originality" that was always partly a fiction — and it certainly isn't a policing regime that fails 94% of the time while quietly making Turnitin and its peers rich.

The honest question was never "did a student use humanity's collective knowledge to help write this paper." Of course they did — so does every professor, every time they open a book. The honest question is who owns that collective knowledge, who got paid two dollars an hour to build it, and whether universities are going to keep punishing students for the enclosure of a commons the institutions themselves helped hand over.