Australia’s Tertiary Education Quality and Standards Agency (TEQSA) says it is “all but impossible” to detect AI cheating.
Still, as generative AI (genAI) expands, universities have the chance to harness its ability to improve learning and assessment – and to teach students how to use it responsibly beyond the lecture theatre and into their working lives.
Accelerating accessible learning
One growing resource is the AI teaching agent Cogniti, developed at the University of Sydney (USyd). Educators across disciplines can tailor Cogniti to their subjects, providing students with new ways to engage with course material. For instance, they can turn their lecture notes into a podcast to listen to while commuting – a great boost to accessibility.
Cogniti doesn’t hand out answers; it challenges students’ thinking. For instance, QUT’s Professor Grant Hamilton has created a chat-based field study guide for his biology students. The interactive conversational agent withholds recommendations until the student reasons through them, turning experimental design into a critical thinking exercise.
Hamilton’s colleague, Associate Professor Jim Hogan, notes that AI is also extremely valuable in providing formative feedback, and makes a brilliant practice agent. “There’s no way we could afford an individual tutor prompting you with questions all the time,” he says, adding that universities would need “an army of people with the right skill level” to achieve the same outcome.
Beyond knowledge: building capability
The University of Queensland’s Professor Jason Lodge, lead author of a new TEQSA resource about quality learning in a genAI future, says many good things about university education don’t need changing – but others do.
“Students still need to have deep disciplinary knowledge to build on – and AI doesn’t remove that need. But we’re finally moving away from being obsessed with the cheating aspect of AI towards a discussion about adaptive capabilities,” says Lodge.
Adaptive capabilities are fundamental to students’ work beyond graduation, and will hold no matter how much AI changes in the future. They include learning to navigate new ideas, thinking critically and asking the right questions, and adapting to new technology.
With USyd bringing in an AI Literacy and Competency unit this year, and rapidly rising numbers of lecturers developing their own AI teaching agents, building these capabilities seems a sure bet.
AI-resilient assessment
The rise of genAI is also driving a re-examination of how student learning is assessed. “We’ve got to look at changes to assessment that make the advantage of using AI go away, and make its use part of the assessment task,” says Hogan.
Lodge agrees, pointing out that it’s important for assessment to demonstrate a student’s genuine understanding, instead of just submitting essays likely written by AI.
“Good old pen-and-paper exams, interactive oral assessments and practical demonstrations all assure learning, which is what TEQSA is ultimately enforcing through its standards framework,” he explains.
Tomorrow’s teaching, today
The capacity for self-directed, active learning, combined with the requirement for students to explain, apply and defend their thinking, makes genAI a potentially powerful learning tool. And with lecturers already using AI themselves, it makes sense that students do too – as long as both use it transparently and with integrity.
– Written by Kath Kovac
Read more
- Discover more TEQSA resources for institutions, staff and students about the potential impacts and benefits of gen AI tools for teaching, learning, assessment and research.
- Find out more about the benefits of Cogniti for educators and students.
