Let's be honest about something first: AI tools are genuinely useful. They can explain a concept at 2am when no tutor is awake, summarise a reading in seconds, and give you ideas when you're staring at a blank page. It makes complete sense that students use them. Anyone pretending otherwise isn't being honest about the pressure of studying.
This guide won't tell you "don't use AI." But it will explain what you're trading when you do, and how to use it without losing your ability to learn.
Here's the core of it: the assignment was never really the point. Nobody needs your essay on first past the post, or your three-page maths proof. The point was what working on the assignment builds in you: the ability to construct an argument, work through problems logically, and find the flaw in your own reasoning. The assignment is like a gym, and thinking and reasoning about the work is the muscle.
But when AI does the work, the assignment gets done, and none of that growth in ability or reasoning does.
This is what's called cognitive offloading: using external tools to reduce the mental effort required to do a task.
What the research is starting to show
A 2025 MIT Media Lab study put EEG headsets on 54 students writing essays over four months, one group used ChatGPT, one used search engines, and one used no tools. The ChatGPT group showed the weakest brain connectivity of the three, reported the lowest sense of ownership over their essays, and, most strikingly, often couldn't accurately quote from essays they had "written" minutes earlier (MIT Media Lab, full paper).
Because we don't cherry-pick, it's important to state that this was one study, with 54 participants, and it's been incredibly over-interpreted online. It doesn't prove AI "rots your brain." But what it does suggest is that when the tool does the thinking, your brain doesn't. Skills you don't practise, you don't keep.
Meanwhile, the suspicion problem is hurting everyone
While that's happening, we're also seeing a growing sense that no one trusts anybody's work anymore.
What happened in NZ
All eight NZ universities adopted Turnitin's AI detection when it launched in 2023. By 2025, it was falling apart. Massey University switched its AI detection off entirely after students reported being falsely accused, with the student association filing a formal report of concern over repeated failures. One student stated: "Many people have been accused, including myself, and it creates very bad anxiety" (Massive Magazine). RNZ reported other NZ universities quietly abandoning the tools too, with some courses returning to in-person, on-paper tests instead (RNZ).
Across the Tasman, it was much more intense. One Australian university logged nearly 6,000 misconduct cases in a year, many based on the detector alone. Students described entire semesters under a cloud, handing over search histories and handwritten notes to prove their innocence (RNZ). Turnitin itself admits a false positive rate that, at scale, means thousands of innocent students flagged, and says its tool "should not be used as the sole basis for adverse actions." It was anyway.
This has resulted in the worst of both worlds. Students using AI heavily often aren't caught, students who write every word get flagged at every turn, and increasingly, students have started recording their screens while writing assignments, just in case they need proof later.
Protect yourself (whether or not you use AI)
- Write in Google Docs or Word with version history on. A document built over days is your best evidence
- Keep your rough notes, drafts, and reading highlights until after marking
- If accused, ask for the specific evidence, and know that a detector score alone is not proof, by the detector company's own admission
- Know your uni's actual AI policy for each course, because it varies, which brings us to the next problem
The rules are different in every classroom
Right now in NZ, AI rules can differ between universities, between courses at the same university, and between two lecturers teaching the same paper. One encourages AI-assisted brainstorming as resourcefulness. Another treats it as misconduct. This can put students in a position where they're left guessing, and the guess can cost them.
That inconsistency isn't your fault, but it is your problem. So here are a few tips to hopefully help you navigate this:
- Check the course guide for every paper, every semester. Don't assume last semester's rules carried over
- If the policy is unclear, ask in writing. An email reply from your lecturer saying "yes, using AI for X is fine" is a lot more reliable and trackable than a vague memory of something said in a lecture
- When allowed to use AI, disclose how you used it if there's any mechanism to. It protects you, and it's honest
- If your uni's policies contradict each other, say so through your class rep or student association. Massey was able to change their AI policy because students advocated for themselves
But ultimately, the longer-term fix is consistency across a university, there needs to be clear expectations as to what's allowed when students use AI. That's a conversation NZ's education system hasn't finished having.
Using AI so you still own the learning
If you're going to use these tools, and most students are, the line that matters is this: AI should make you think more, not less.
Here are a few ways we suggest you use AI that give you a much better chance of learning:
- Explaining a concept you're stuck on, then closing the tab and writing your own version
- Quizzing you on material before an exam
- Critiquing a draft you already wrote: "what's the weakest part of this argument?"
- Generating counterarguments for you to wrestle with
- Unblocking a blank page with questions, not paragraphs
And here are some things we suggest you avoid, which have a much greater chance of degrading your abilities:
- Generating the draft you then lightly edit
- Summarising readings you never open
- Answering the question before you've attempted it
- Anything where you couldn't explain the submitted work out loud, unaided
If you have any doubt about which bucket your use falls into, ask yourself this: could I defend this work in a conversation with no notes? If yes, the AI was a tool. If not, it was a substitute, and the person it substituted was you.
What we recommend most of all is to check your relationship with the tool.
If you start to notice the reflex forming to just ask AI first across anything, not just assignments, whether that's medical questions, money decisions, or arguments with your flatmate, take note. Each individual use may be reasonable. But not if it's your immediate, consistent go-to.
If you're worried you fall into this, a rough self-check:
- When did you last sit with a hard problem for more than ten minutes before reaching for a tool?
- Can you still write a first draft of anything, cold?
- When AI gives you an answer, do you ever check it, or has its confidence become your confidence?
The goal isn't to never use AI. But it's to stay the kind of person who can think without the tool, and who uses it by choice rather than dependence. Your degree is supposed to certify that you've learnt something, that you've worked hard, that the hours late at night meant you can do what that paper says, maybe even more. Make sure that's true.