Australian Students Are Asking the Right Question About AI and NAPLAN

Australian Students Are Asking the Right Question About AI and NAPLAN

When students from Northern Beaches Christian School wrote to Australia’s Education Minister calling for AI literacy to be embedded in the curriculum – and for assessments such as NAPLAN to be reconsidered – it would have been easy to dismiss their argument.

Some adults may hear:

“Students want to use ChatGPT.”

Or:

“They want easier exams.”

I heard something different.

I heard young people asking education leaders to stop pretending that AI is a temporary disruption.

They are asking a more important question:

If AI will be part of how we learn, work and make decisions, what should schools teach – and what should they assess?

I am approaching this question from two perspectives.

I am a father raising a daughter in Australia.

I am also a Principal AI Architect who works with enterprise AI systems, including generative AI and AI agents.

From both perspectives, I believe the students are right to challenge the status quo.

But I do not believe the answer is to abandon foundational knowledge or remove every traditional assessment.

The better answer is to build a new assessment model for an AI-enabled world.

AI does not make literacy and numeracy less important

There is a dangerous idea emerging in some conversations about AI:

If a machine can write, calculate and retrieve information, children no longer need to master those skills themselves.

I believe the opposite is true.

AI can generate fluent but incorrect answers. It can fabricate sources, misunderstand context and produce convincing explanations built on faulty reasoning.

A child with weak reading skills may not recognise that a source is unreliable.

A child without number sense may accept an incorrect calculation because it appears professionally presented.

A student with little background knowledge may not know what questions to ask or which assumptions to challenge.

AI is most useful to people who already possess enough knowledge and judgment to evaluate its output.

That means literacy, numeracy and subject knowledge remain essential.

They are not outdated skills.

They are the foundation that allows a person to use AI safely and intelligently.

This is why I would not simply abolish NAPLAN.

Australia still needs a national view of whether students are developing fundamental reading, writing, language and numeracy capabilities.

The limitation is not that NAPLAN measures foundational skills.

The limitation is that foundational skills are no longer the complete picture of readiness for the future.

Schools cannot pretend AI does not exist

At the same time, simply banning AI is not a sustainable education strategy.

AI is already part of university, software development, finance, healthcare, marketing, design and professional services.

By the time today’s primary school students enter the workforce, using AI assistance will likely be normal in many occupations.

A school system that teaches students only how to avoid AI does not prepare them for that reality.

It may simply teach them to use AI secretly.

Real AI literacy is not the ability to write a clever prompt.

It includes understanding:

• why AI can be wrong;
• • how to verify claims and sources;
• • how bias can enter a system;
• • how to protect personal information;
• • how synthetic media and deepfakes can manipulate people;
• • when AI assistance is appropriate;
• • how to disclose its use honestly;
• • and how to keep human judgment in control.

Children will not reliably develop these capabilities by experimenting alone with public chatbots.

They need structured teaching, appropriate tools and teachers who are themselves supported and trained.

We should assess the thinking, not only the final answer

Generative AI exposes a weakness that already existed in education:

A polished final submission does not necessarily prove genuine understanding.

A student can produce an impressive essay with AI.

But can the student explain the argument?

Can they defend the evidence?

Can they identify where the AI made a mistake?

Can they respond when the assumptions change?

Can they transfer the idea to a different problem?

These questions point toward a more balanced assessment system.

1. Protected foundational assessments

Some tasks should remain AI-free.

Students need opportunities to demonstrate that they can independently read, write, calculate, recall key knowledge and reason through age-appropriate problems.

Without this foundation, responsible AI use is impossible.

2. AI-permitted assignments with disclosure

Other tasks should allow AI, because that reflects how work increasingly happens outside school.

But students should document:

• which tool they used;
• • the prompts or instructions they provided;
• • what the system generated;
• • which suggestions they rejected;
• • what they changed;
• • and why the final judgment remains their own.

This changes AI use from hidden assistance into an observable learning process.

3. Oral explanation and defence

Teachers can ask students to explain their work:

Why did you choose this evidence?

Which source was most trustworthy?

Where was the AI response weak?

What would change if one condition were different?

A student who understands the work should be able to discuss it.

A student who has outsourced the thinking will struggle.

4. Authentic, real-world tasks

Schools should increasingly assess capabilities that matter beyond exams:

• solving unfamiliar problems;
• • evaluating competing sources;
• • working with others;
• • creating something original;
• • communicating decisions;
• • and using technology responsibly.

These are not “soft” alternatives to academic knowledge.

They are the capabilities that turn knowledge into useful action.

The parent perspective

As a parent, I do not want my daughter to become dependent on AI.

I want her to be able to read, write, calculate, explain and think without it.

But I also do not want her education to prepare her for a world that no longer exists.

I want her to understand AI well enough to use it, question it and refuse its advice when necessary.

That balance matters.

The goal should not be to produce children who never use AI.

Nor should it be to produce children who ask AI to do every difficult task.

The goal should be to develop people who can think independently and use AI deliberately.

The architect perspective

In enterprise AI, we do not place an advanced model into production and assume everything will work.

We define boundaries.

We verify outputs.

We control access to data.

We keep audit trails.

We use human review for high-risk decisions.

We monitor whether the system is producing the intended outcome.

Education AI requires the same discipline.

The question is not merely whether students or teachers can access AI.

The questions are:

Is the tool appropriate for the child’s age?

Is the content aligned with the curriculum?

What student information is being processed?

Who remains accountable for the decision?

Does the system improve learning – or simply produce more content faster?

Technology without governance does not create better education.

It creates faster uncertainty.

The students are not asking to stop learning

The most important part of this story is that the students are not arguing that AI should do everything for them.

They are asking adults to acknowledge the environment they are growing into.

Their message is essentially:

Do not pretend AI does not exist.

Teach us how to use it responsibly.

And assess the capabilities that will genuinely matter in our future.

That is a reasonable request.

My position is therefore not “remove NAPLAN” or “ban AI.”

It is:

Protect foundational knowledge. Teach AI literacy explicitly. Redesign assessment so students must show their reasoning, judgment and responsible use of technology.

The progression should be:

Foundational knowledge → critical thinking → AI literacy → responsible problem-solving

AI should not replace thinking.

It should amplify people who have learned how to think.