Every semester, someone in a staff meeting asks the same question: how do we stop students using AI on this? It’s a fair question, but it’s the wrong one to build a course around.
Detection tools are unreliable. Generative AI improves faster than any assessment redesign cycle. And our graduates are heading into workplaces where using AI well is already part of the job. A nurse, an engineer or a policy analyst who can’t work alongside these tools will be at a disadvantage.
So the better question is this: where should students use AI, and where must the learning come from them?
Why “AI-proof” is a trap
An essay that resisted AI in 2023 might be trivial for AI today. Reflective tasks, personal narratives and even “cite the lecture” tasks have all been worked around. Chasing AI-proofness turns assessment design into an arms race that educators can’t win and students don’t learn from.
Worse, it pushes us toward banning a tool rather than teaching its use. Students then use it anyway, quietly and badly, with no guidance on judgement, verification or ethics.
The two-lane idea
Many Australian universities are now moving toward what’s often called a “two-lane” approach, and TEQSA’s guidance on assessment reform points in the same direction. The idea is simple:
- Secure lane: assessments that verify what a student can do themselves, under conditions where AI use is controlled. Think supervised exams, observed practicals, oral defences and in-class tasks.
- Open lane: assessments where AI use is permitted or even expected, and the focus shifts to how well students use it.
The two-lane model works well at the level of a whole program. But most of us design learning one activity at a time. That’s where hybrid activities come in.
What a hybrid activity is
A hybrid activity brings both lanes into a single learning experience. It’s a short sequence of stages, and each stage is clearly labelled:
- AI-integrated stages where students use AI on purpose: to generate a first draft, simulate a patient, summarise literature or offer critique.
- AI-resistant stages where the learning must visibly come from the student: an in-class discussion, a live explanation, a decision tied to their placement, or a short oral check.
The resistant stages aren’t there to punish AI use. They act as checkpoints that confirm the student understands what the AI helped them produce. If a student has leaned on AI without learning anything, the checkpoint shows it. If they’ve used it well, the checkpoint gives them a chance to show that too.
Why this works better
It’s honest. Students know exactly when AI is allowed and why. Clear expectations reduce accidental misconduct far more than a blanket ban.
It teaches judgement. The most valuable AI skill isn’t prompting. It’s knowing when an output is wrong, incomplete or unsafe. Hybrid activities make that judgement the thing being practised.
It’s more durable. Because the resistant stages rely on presence, process and personal context, they hold up better as AI tools improve. Nothing is truly AI-proof, but a live conversation about your own reasoning is a high bar.
It mirrors professional practice. In health, law, business and engineering, professionals increasingly draft with AI and then verify, adapt and take responsibility. Hybrid activities rehearse exactly that.
A quick example
In a second-year nursing unit, students ask an AI tool to draft a care plan for a case study patient. That’s the AI-integrated stage. In the next tutorial, working in pairs and without devices, they mark up the plan: what’s unsafe, what’s missing and what doesn’t fit this patient. That’s the AI-resistant stage. Finally, each student writes a short reflection on what they changed and why.
The assessment no longer rewards the care plan alone. It rewards the student’s ability to evaluate one, which is the skill that matters on the ward.
Where to start
You don’t need to redesign a whole unit. Pick one existing activity and ask three questions:
- Which part of this task would AI genuinely help students learn from?
- Which part must show the student’s own understanding?
- How will I see the difference?
Answer those, and you’re already designing hybrid activities.
Our Hybrid Activity Designer builds these staged activities for you, with each stage labelled and explained.