Out of 200 undergraduates surveyed at the University of Ibadan, 170 of them, that is 85 percent, agreed that most students do not understand the ethical implications of using AI in their academic work.
Read that again. Not that students refuse to follow the rules. That they do not know what the rules are.
The study was published on 31 July 2026 in Frontiers in Education, so this is current. And it lands in the same month that something else quietly changed about your final year project.
The thing that actually changed in 2026
Most conversations about AI for your final year project are stuck on one question. Will the detector catch me?
That was always the wrong question, and now it is out of date.
Under the Nigeria Education Repository and Databank, your project no longer ends its life on a shelf in the departmental library. It goes into a national digital store that supports plagiarism checking and can be pulled up by employers and embassies years after you graduate.
We wrote about that in full when NYSC tied mobilisation to NERD clearance.
Here is why it matters for this conversation. A detector is a gate you pass once. A permanent record is different. It sits there, checkable, for as long as the database exists.
The risk stopped being about submission day. It became about the rest of your career.
Nobody actually told students the rules
The Ibadan findings are worth sitting with, because they explain why so many students get this wrong without meaning to.
Alongside that 85 percent, 138 respondents, about 69 percent, agreed that the lack of clear institutional guidelines creates confusion about what counts as acceptable AI use.
At the same time, 170 of them, again 85 percent, said lecturers discourage or frown upon AI use in coursework.
So put those two together. Lecturers signal disapproval. The institution has not written down what is allowed. The student is left guessing.
That is not a discipline problem. That is a policy vacuum, and students are the ones carrying the risk for it.
Where the line actually sits when you use AI for your final year project
Until your department publishes something specific, use this as your working rule.
AI may help you with the work around the thinking. It may not do the thinking.
Your project is supposed to evidence that you can frame a problem, gather data, interpret it and defend a conclusion. That is the thing being assessed. Everything else is scaffolding.
So the question is never whether you touched AI. It is whether the ideas and the analysis are yours.
What AI is genuinely good for on a project
Used this way, AI for your final year project is defensible in almost any viva, and we work like this ourselves.
Finding your way into a literature. Ask it to explain a theory you keep bouncing off, or to suggest what search terms a field actually uses. Then go and read the real papers.
Interrogating your own argument. Paste your paragraph and ask what a hostile examiner would attack. This one is underused and it is the most valuable.
Cleaning up your English. Grammar, sentence length, clumsy phrasing. Especially useful if English is your second or third language.
Explaining your own statistics back to you. If you ran a regression and cannot say in plain words what it means, you are not ready for defence. Use AI to rehearse the explanation until you own it.
What you must not hand over
Your research questions. If AI chose them, you will not be able to defend why they matter.
Your analysis and interpretation. This is the assessed part. Outsourcing it is the actual misconduct.
Your data. Never let a model invent numbers, and never paste raw participant data into a public chatbot. That second one is a data protection problem as much as an integrity one.
Your citations. AI still fabricates references that look perfect and do not exist. Every single source must be opened and confirmed. A fake reference in a permanent national repository is not a small thing.
The permanence test
Before you submit, ask yourself one question.
If somebody pulled this project out of a federal database in 2031 and sat me down to discuss it, could I defend every page as my own reasoning?
If yes, you are fine, regardless of which tools you used along the way.
If no, the problem was never the detector. Fix it now while it is still cheap to fix.
That test is a better guide to using AI for your final year project than any word count of machine-written text, and it survives whatever the detection technology does next.
If you have already leaned on AI too heavily
Plenty of people reading this are halfway through a project they do not fully understand. Abeg, no panic.
Go chapter by chapter and mark any claim you could not explain out loud to a course mate. Those are your weak points, and they are also exactly where your supervisor will push.
Then rebuild those sections from the sources themselves. Read the paper, take your own notes, write the paragraph in your own words.
It is slower. It is also the only version of the work that will survive a viva, and now, a permanent record.
One more honest note. The Ibadan study found 177 respondents, 88.5 percent, agreeing that dependence on AI tools breeds academic laziness. Students themselves are saying it. The tool is not the enemy, but the dependence is real and worth watching in yourself.
What we tell the students we work with
Our position on AI for your final year project has not changed because the technology changed. Use it as a tool you direct, not one that directs you.
If you would like more on the practical side, we have written about using AI for assignments without switching off your brain, how to study with AI without getting flagged, and why detectors keep falsely flagging Nigerian students. If you are at postgraduate level, start with finishing your dissertation.
And if you are staring at a project that now has to last forever and you are not sure it is good enough, that is the exact work we do. Send us what you have and we will go through it with you.
Your project used to be a hurdle. It is a record now. Write it like something you will still be happy to put your name to in ten years.