Schools are responding to AI in assessment in a handful of ways. Some ban it, some try to detect it, some redesign their tasks and some bring more assessment under supervision. Others are still deciding. Each response addresses part of the problem, and each carries a cost.
The central task for teacher and student is to establish what the student understands. My view is that our response to AI should start there.
Why does every school have to answer this now?
A finished assignment can now contain a mixture of student thinking, permitted AI assistance and generated material that the student barely understands. The quality of the submission alone cannot reliably distinguish those contributions.
That creates a problem for reporting. Families need to know that a result reflects their child's capability. Students need consistent expectations about what help is allowed and what they must demonstrate independently.
Teachers also need a workable way to make that judgement. If every polished submission becomes an investigation, the school has added a substantial new demand to marking. Asking teachers to become AI detectives is a poor use of their expertise.
What are schools actually choosing?
Bans draw a boundary around acceptable assistance. Detection software attempts to identify generated text after submission. Task redesign changes what students are asked to do, often adding personal application, checkpoints or explanations of their reasoning.
Supervision changes the conditions in which the evidence is produced. Canterbury College in Queensland has announced that assessment contributing to reported results in Prep to Year 10 will be fully supervised from 2027. Take-home assignments will cease to count towards those results. Its approach includes creative responses and extended writing completed in classroom conditions, alongside examinations.
The College also plans to increase its investment in AI education. Its decision shows that teaching students to use AI and requiring independent demonstrations of learning can sit within the same school policy. Years 11 and 12 remain subject to QCAA syllabus requirements.
Waiting is another response, whether through a deliberate review or an unresolved decision. In the meantime, individual teachers still have to decide how to handle the work in front of them.
What does each response cost?
A ban gives students a clear rule, but enforcing it beyond environments a school controls is difficult. Schools also need to explain which uses are prohibited and where AI can legitimately support learning. A blanket instruction can leave those distinctions unclear.
Detection adds software costs and still requires judgement. QCAA's guidance for schools warns that AI detection tools are inconsistent and should be used cautiously. A score alone is an insufficient basis for deciding who produced a piece of work. Someone still has to examine the concern and speak with the student.
Task redesign takes planning time. Asking students to apply knowledge to a particular context can improve the task, but it cannot make outside assistance impossible. The design also needs to include opportunities for the teacher to see the student's understanding develop.
Supervision provides stronger assurance about the conditions in which work is completed. It asks for classroom time, suitable spaces and consistent administration. It can accommodate extended work across several lessons. Where a school instead concentrates assessment into short, timed tasks, it needs to consider the effects on inquiry, revision and students who need adjustments.
Waiting carries a workload of its own. Without an agreed approach, teachers must keep resolving uncertainty individually, and students can encounter different expectations across subjects.
Does secure have to mean supervised?
Secure assessment can draw on supervised work, evidence of how a response developed and discussion with the student. The aim is to establish what the student understands and can demonstrate independently.
For me, the Socratic method is the gold standard when understanding really counts. A teacher asks a student to explain their reasoning, then uses the answer to decide what to ask next. That exchange can reveal gaps and give the student an opportunity to demonstrate the depth of their understanding.
The constraint is time. Schools do not have the time or resources for a lengthy oral defence of every assessment. The practical challenge is to make this kind of focused discussion easier for teachers to use when it will add most value.
QCAA's Prep to Year 10 authentication guidance includes observing work at different stages, consulting students at checkpoints and interviewing them about their responses. These give teachers several ways to examine the relationship between the submission and the student's understanding.
Provenance, meaning the available record of how work developed, can provide the context for that discussion. Drafts and revisions can help a teacher decide what to explore and prepare questions linked to the assessment criteria.
A substantial change between drafts, for example, gives a teacher a starting point: what changed your argument, and why? The student's answer can guide the next question about the evidence or the reasoning. This is Socratic questioning grounded in the student's own work.
Provenance and dialogue provide evidence to interpret. A record of revisions cannot prove authorship, and teachers need to consider a student's communication needs when questioning them. The judgement rests on the evidence gathered across the task.
Where are we heading?
Schools need an approach that reflects the subject, year level and what students are being asked to demonstrate. Technology can make that easier by bringing the relevant history into view and helping teachers prepare useful questions for a focused discussion.
At rubric+, we already help teachers capture and retain assessment evidence at descriptor and question level.
We are building on that foundation to make reviewing student work easier and support more focused conversations about understanding. By visualising the provenance of how the assessment was created, mapping out the natural drafting timeline, version milestones and structural evolution of the work, we provide transparent insights into the student’s learning journey.
So now teachers can continue to decide what the evidence means and which questions to ask.
Whatever approach your school chooses, it needs to give teachers a practical way to judge what students understand and confidence in the results they report. That means deciding where supervision is necessary and how evidence gathered during extended tasks can support that judgement.
So, what is your response?
About rubric+
rubric+ is an assessment evidence platform built for Australian schools. Designed by educators, it preserves the professional judgement teachers make during marking and structures it into evidence every student, teacher, leader, and parent can use. From descriptor-level performance insights to reporting that writes itself, rubric+ makes the invisible work of teaching persistent, visible, and actionable.