_edited.png)
NSW Wants to Take Year 12 Assignments Out of the Home. Is AI Changing School Forever?
- Madison | Nudge Your Career
- 2 days ago
- 6 min read
ChatGPT was supposed to be a tool. Now NSW is asking whether it has made some school assessments impossible to trust.
There was a time when being given a Year 12 assignment meant heading home, opening a laptop, searching Google, reading a few articles, writing a few thousand words and probably procrastinating somewhere in between.
Now there is another option.
Open ChatGPT.
Within seconds, students can generate an essay, restructure an argument, explain a complex topic, create a study plan or rewrite something they have already written.
And that has created a very uncomfortable question for the NSW education system:
If a student can submit work created by AI, how do we know what the student actually knows?
The NSW Government is now taking that question seriously.
NSW wants to change take-home assessments
On 10 August 2026, NSW Deputy Premier and Education Minister Prue Car asked the NSW Education Standards Authority (NESA) to urgently examine the impact of artificial intelligence on student learning and HSC assessment.
One of the immediate measures being considered is a moratorium on unsupervised, take-home assessment tasks while the broader review takes place.
In other words, the proposal isn't simply:
"Ban ChatGPT."
It is potentially much bigger.
NSW is considering changing where and how students complete assessment work.
Instead of completing certain high-stakes assignments at home, students could increasingly be required to complete them under supervised conditions at school.
And this could begin affecting students moving into Year 12 as early as next term, depending on NESA's recommendations.
Why is this happening?
Because AI has changed the meaning of a take-home assignment.
A teacher might set an essay because they want to assess a student's:
research ability
writing ability
critical thinking
understanding of the subject
ability to construct an argument
ability to analyse information.
But if a student asks an AI tool to produce the essay, the final document might demonstrate the capabilities of the AI rather than the student.
That's the problem.
The government says the integrity of the HSC could be undermined if assessment outputs can be produced without students performing the intellectual work required by the course.
And this isn't just theoretical.
NSW recorded more than 1,000 cases of cheating in HSC school-based assessments in 2025, according to reporting surrounding the government's announcement.
But ChatGPT isn't automatically cheating
This is an important distinction.
NESA already has rules governing the use of AI in HSC and RoSA assessments.
Students must follow the assessment conditions set by their school and NESA, and unauthorised AI use can constitute academic malpractice.
But AI itself isn't necessarily the enemy.
There is a huge difference between:
"Explain this concept to me."
and
"Write my entire assignment."
There is also a difference between using AI to brainstorm ideas and submitting AI-generated work as your own.
NESA recognises that AI can potentially be used to augment teaching and learning, while also emphasising that it cannot replace strong curriculum, assessment and teaching.
So the real issue isn't simply whether students are using ChatGPT.
It's how they're using it.
The problem with trying to catch AI
Schools have already been trying to work out how to identify AI-generated work.
But this creates another problem.
AI detection isn't a magic solution.
A piece of writing can be edited after being generated by AI.
AI can be used to help someone develop an argument before they write the final version themselves.
And a student's legitimate writing can potentially be mistaken for AI-generated content.
That makes the question of proving authorship extremely complicated.
Some school assessment policies already require students suspected of using AI to provide evidence such as drafts, timestamps, process diaries or other material showing how their work was developed.
But that creates an enormous administrative burden for teachers.
Imagine having to investigate hundreds of assignments to determine whether each student actually wrote their own work.
The NSW Government's response is essentially:
Maybe we shouldn't have to detect our way out of this.
Maybe some assessments simply need to happen under supervision.
Half of an HSC mark makes this serious
For most HSC courses, school-based assessment contributes 50% of a student's final HSC mark. School-based assessments can include tests, written assignments, practical activities, fieldwork and projects.
That means this isn't just about whether someone used ChatGPT to polish a homework assignment.
These assessments can materially affect a student's final result.
And HSC results can influence university admission and future opportunities.
So the government has a legitimate question:
Can an assessment be considered fair if some students complete it independently while others use AI to do the intellectual heavy lifting?
But here's where Nudge thinks the conversation gets interesting
We shouldn't pretend AI doesn't exist.
Because students are about to enter a workforce where AI will be everywhere.
The workplace isn't going to say:
"Congratulations on finishing Year 12. Now please forget everything you learned about artificial intelligence."
Quite the opposite.
Employers are increasingly going to expect young professionals to understand how to use AI.
They will need to know how to:
write effective prompts
check AI-generated information
identify hallucinations and inaccuracies
protect confidential information
use AI without compromising intellectual property
improve productivity
critically evaluate AI-generated work
know when a human needs to take over.
Those are career skills.
So banning AI entirely isn't necessarily the answer.
Perhaps we need to teach AI literacy instead
There is a bigger opportunity here.
Instead of treating ChatGPT as something students either secretly use or aren't allowed to touch, schools could teach students how to use AI responsibly.
Imagine an assessment where students are explicitly allowed to use AI.
But they have to submit:
1. Their original thinking
2. The prompts they used
3. The AI responses
4. Their critique of those responses
5. What they changed and why
6. Their final work
Suddenly, AI becomes part of the assessment rather than something students are hiding from teachers.
And the student is still required to demonstrate critical thinking.
In fact, that could arguably be a more realistic assessment of the world students are about to enter.
But some things probably should remain supervised
There is also a strong argument for keeping certain assessments AI-free.
If you're trying to establish whether a student can independently write, analyse, calculate, recall or reason, supervised assessment makes sense.
The question isn't whether students should ever be separated from technology.
It's whether every assessment should measure the same thing.
A timed classroom task might measure independent knowledge.
A take-home project might measure research and project management.
An AI-assisted assignment could measure technology literacy and critical evaluation.
Different assessments can test different capabilities.
And what about students who struggle with exams?
This is where the proposed changes need to be handled carefully.
Take-home assessments can provide flexibility for students who don't perform well under traditional exam conditions.
They can give students more time to formulate ideas, conduct research and demonstrate what they know.
Moving more assessment into supervised environments could therefore create new challenges for some students.
The solution can't simply be:
"AI is difficult to control, so everyone sits in a room and writes an exam."
Education is supposed to measure learning, not simply someone's ability to perform under pressure.
Any changes will need to consider reasonable adjustments, accessibility and the different ways students demonstrate knowledge.
The uncomfortable truth
ChatGPT hasn't broken education.
It has exposed some of the weaknesses that were already sitting underneath it.
If an assignment can be completed by asking an AI to produce a polished response, perhaps the assignment wasn't measuring the right things in the first place.
And if a student can receive a higher mark simply because they know how to get an AI system to produce better work, we've got a different kind of inequality emerging.
The answer isn't necessarily to ban technology.
It's to make sure the technology isn't doing the learning for the student.
From the classroom to the workplace
This matters beyond the HSC.
The students affected by these changes will eventually become graduates, employees, managers, entrepreneurs and leaders.
They will be expected to work alongside AI.
They will need to know when to trust it.
When to challenge it.
When to disclose its use.
And, most importantly, when not to use it.
That's exactly why AI literacy should become part of career education.
Because the future workplace won't reward someone simply for knowing how to avoid AI.
It will reward people who know how to use it without losing their own judgement.
So, should Year 12 students be banned from using ChatGPT?
Not necessarily.
But perhaps we should stop pretending that every assignment can remain exactly the same now that AI exists.
NSW's proposal is really a test of something much bigger than ChatGPT.
It's asking whether our education system is still assessing students, or the quality of the tools they have access to.
The answer shouldn't be to fear AI.
It should be to teach young people how to use it responsibly, ethically and intelligently.
Because by the time today's Year 12 students enter the workforce, AI won't be the new technology anymore.
It will simply be part of the job.
Don't teach young people to fear AI. Teach them to think before they use it.
The most valuable skill in an AI-powered workplace won't be knowing how to get a chatbot to write for you.
It will be knowing what to ask, what to challenge, what to verify and when to put the chatbot away and think for yourself.
Comments