1 hour ago
3
June 27, 2026 — 5:00am
University mid-year exams are here, but most of my fellow students of 2026 can be reassured that their futures will not be on the line from a few hours spent in an open hall like the Royal Exhibition Building. Many subjects don’t have exams at all or have “take-home exams”. For students who do have to sit exams, many more will have already secured most of the grades from their bedrooms without having sat a single in-person test.
Completing work at home isn’t inherently a problem. What alarms me is the flagrant, unregulated way that I’m seeing my generation of uni students use artificial intelligence to do the work for them. Students aren’t using AI to think better; they’re using it to avoid thinking at all. Universities say they are discouraging this kind of behaviour, while structuring student assessments in ways that promote it. It’s no wonder academics like Kylie Moore-Gilbert and Marvin Starominski-Uehara are questioning the value of a degree.
I’m no technophobe who believes that ChatGPT is the devil incarnate and that we should all write exams with ballpoint pens. In fact, I’m a dyslexic, first-year arts student who spent the end of year 12 asking ChatGPT for sample exam questions and model answers, which is a useful and legitimate study tool. But I’m frustrated at the unfair way AI is being used.
Every week, I attend tutorials. My experience: the tutor asks a question, kids consult ChatGPT to provide a script for answers; the things we allegedly come to university to hone. Meanwhile, the tutor looks around bleakly at a sea of blank faces behind laptops.
Privately, students discuss which parts of their essays ChatGPT wrote for them, or how it, or Grammarly, or some other AI “proofread” their work and “made it sound more academic”. (In case you aren’t a ChatGPT user and aren’t aware, this means ChatGPT takes your essay, completely changes it – “keeping your sentiment,” of course — and spits you out a “polished, HD-level, final version”.)
It’s not just us arts students. In the University of Melbourne subjects Biomolecules and Cells, or Fundamentals of Chemistry, for example, 40 per cent of students’ grades are comprised of out-of-class assessment. By which I mean an online worksheet completed from bed. That might have been fine six years ago in the grip of COVID-19. But today, when ChatGPT can complete any calculation or short answer question in an instant, and students in tutorials gloat about their “easy 100 per cent. No study,” (and, yes, this is a quote) and “I just chucked it all into Chat,” I feel like we might have a problem.
When AI has secured you top marks for your take-home work, you don’t need to do much on your exam to get a pass. And even in your exam, AI can do the work for you. When a cohort is told that their biology end-of-year exam allows a double-sided cheat sheet, I’ll give you a clue what the immediate response is: “Chat-GPT, here’s my entire bio syllabus and the past exams … could you summarise the whole thing for me, highlighting important information?”
And I don’t want to single out Melbourne Uni. A Monash University law student told me of his recent video assignment for, ironically, Contracts Law: “Yeah, just Chattersed the whole thing, mate. The f--- would they know; it’s a video.”
Universities are where we students should be developing our values and views through reading, listening and conversation. So, what happens when generated summaries replace listening, reading culture withers, and conversation lapses into silent video watching?
Any of us would be concerned to discover our doctor didn’t watch lectures in medical school but instead used Claude AI software to summarise already AI-generated transcripts. Yet this is exactly what students – at UTAS, Monash, Uni Melb, USYD, UNSW, and UQ, to name a few – have privately admitted to doing, either to me in person when I researched this article, or on Reddit threads. Monash University even promotes the use of AI in this way with an instruction manual. We would be just as worried that universities are graduating schoolteachers who never learned what they are to teach. Is the next generation of teachers going to pride themselves most of all for the AI prompts they pass on to their students?
With their lack of enforcement regarding AI use, universities all but condone it. Not least because their detection software is, according to academics, entirely ineffectual.What happens if a student is caught using AI by the official Turnitin software that almost all universities use? At my university we have been told the student would need to send a voice memo sometime in the following week “explaining their authorial intent”. “Hey, Chat-GPT, write me a script about my authorial intent.” Perhaps not the most effective deterrent.
There are three problems (and no, Chat-GPT didn’t provide them).
First, universities charge exorbitant fees and receive billions in taxpayer money, and as such, have a moral obligation to educate their students; for the sake of both students and society.
Second, an MIT study found large language models (LLMs), like ChatGPT, truly rot the brain’s neural networks, with users exhibiting “up to 55 per cent reduced connectivity … suggesting AI assistance fundamentally restructures our cognitive architecture.” This leads to “cognitive debt,” where “LLMs replace… independent thinking”. That’s a lot for a generation to give up.
We face “diminished critical inquiry, increased vulnerability to manipulation, [and] decreased creativity,” the study warns. “[We] not only forfeit ownership of the ideas but also risk internalising shallow or biased perspectives … influenced by the priorities of the LLM’s shareholders.”
Finally, no one knows what the wider consequences of this will be. Sure, AI can help streamline mundane tasks like making spreadsheets. But AI can’t think. And at this rate, neither will humans. We are future employees, future teachers, doctors, lawyers, engineers, biologists, politicians, pathologists, economists, architects, writers, and researchers. Society will suffer if we can’t think, innovate, make conscious decisions, communicate humanely or reason.
Universities are purportedly full of the smartest people in the world. They should be able to ensure that tasks assess students’ capability to think. Universities already selectively use Cadmus, software that tracks absolutely everything that happens in a document, preventing copying, pasting and sharing. So why not implement it for everyone? Profit and rankings are important; integrity less so.
If AI does take over the world, it won’t be because “Skynet” blows us all to hell. (If you don’t know what Skynet is, ask Chat.) It will be because we take for granted that the raggedly creative beings who came before us have already done all the tests and neatly packaged up all the answers in an AI, and we blithely accept submission, forgetting our unique ability: conscious thought.
Samuel Castle is a student at The University of Melbourne.
The Opinion newsletter is a weekly wrap of views that will challenge, champion and inform your own. Sign up here.
