Catch them or trust them? The good cop, bad cop dilemma facing unis
With AI now appearing in more than half of Australian tertiary submissions, a CQUniversity educational neuroscience expert says universities need to stop asking whether students used AI and start asking whether they can demonstrate what they learned.
Professor Ken Purnell, Head of Educational Neuroscience at CQUniversity, has released updated AI guidelines for his students, arguing that academic integrity in the AI era should be built around trust, transparency and authentic assessment, rather than surveillance and suspicion.
Turnitin reported in July that 53.6 per cent of Australian tertiary submissions processed through its system contained some level of AI-generated content between October 2025 and April 2026. Ten per cent contained more than 80 per cent AI-written content.
Professor Purnell said the data showed universities needed to move beyond simply asking whether students had used AI.
“The question shouldn't simply be ‘Did you use AI?’ It should be ‘Can you explain, defend and take responsibility for your work?’”
Rather than requiring students to document every interaction with an AI tool, Professor Purnell advocates a simple approach: acknowledge AI when it has meaningfully shaped ideas, structure or wording, while retaining responsibility for the accuracy, analysis and conclusions.
He said creating an environment where students could be honest about AI use was just as important as setting clear rules.
“Academic integrity shouldn't mean proving you didn't use AI. It should mean being able to demonstrate that you understand what you've submitted.”
The approach reflects a broader shift in higher education, with Australia's Tertiary Education Quality and Standards Agency (TEQSA) increasingly emphasising assessment design that demonstrates genuine learning in an AI-enabled environment.
Professor Purnell said assessment needed to focus less on the final polished product and more on the student's thinking.
“If AI can produce a polished essay, then a polished essay alone becomes less useful as evidence of learning,” he explained.
“We need assessment that makes the student's thinking visible – their reasoning, their judgement, their ability to apply knowledge and explain why they reached a particular conclusion.”
He said this could mean greater use of reflective work, staged projects, real-world problems, presentations and oral discussions where students explain and defend their ideas.
Professor Purnell said this did not mean treating AI as the enemy.
Emerging research suggests AI can support learning when used appropriately, but concerns remain about cognitive offloading when students allow technology to perform the thinking they are supposed to develop themselves.
“There is a huge difference between using AI to challenge your thinking, explore an unfamiliar concept or get feedback, and asking AI to do the thinking for you,” Professor Purnell explained.
“The goal isn't to produce graduates who never use AI. It's to produce graduates who know when to use it, when not to use it, and who remain capable of thinking when the machine gets it wrong.”
