AI cheating isn’t breaking universities, but old-school assessment might
Claims that artificial intelligence is fuelling widespread cheating and eroding the value of university degrees are overstated – and risk distracting from the real challenge facing higher education, according to CQUniversity educational neuroscience expert Professor Ken Purnell.
Responding to recent media reports that suggest students are using AI to “cheat their way to a degree”, Professor Purnell said the evidence does not support a surge in academic dishonesty. Instead, it highlights how poorly many assessments are designed for an AI-enabled world.
“Banning AI in assessment might sound tough, but it’s largely unenforceable,” Professor Purnell said.
“If universities want to protect the value of their degrees, they need to redesign assessment – not declare a war they can’t win.”
Large-scale data paints a more measured picture than anecdotal student accounts, he said.
Independent analysis of millions of submissions by Turnitin has found only a small minority are predominantly AI-generated, with rates remaining largely stable since generative AI tools became widely available.
Longitudinal student surveys also show cheating behaviours have not increased beyond historical norms.
“This isn’t an integrity crisis,” Professor Purnell said. “It’s a design problem.”
According to Professor Purnell, assessments that rely on single, take-home written submissions are inherently vulnerable – regardless of whether AI exists. Simply adding a line that says “no AI allowed” does little to prevent misuse.
“The research is clear: prohibition doesn’t work,” he said. “What does work is structural change – staged assessments, in-class components, process-based tasks, and evaluation that focuses on how students think, not just what they produce.”
Importantly, Professor Purnell said evidence also shows that when generative AI is integrated purposefully into teaching, it can improve learning outcomes rather than undermine them.
“When students are taught how to use AI critically and transparently, learning improves,” he said.
“The danger isn’t AI itself – it’s pretending higher education can keep operating as if nothing has changed.
“Degrees don’t lose value because students use AI,” he said. “They lose value when institutions cling to outdated assessment models that no longer reflect how knowledge, work and decision-making operate in the real world.”
He said the debate needs to move beyond fear-driven headlines and towards evidence-based reform.
