The University of Tasmania has become the latest Australian institution to concede that generative artificial intelligence has outrun the way it tests students, announcing it will strengthen oversight of assessment to protect the credibility of the degrees it awards. The move, reported by ABC News, lands at a moment when vice-chancellors across the sector are grappling with the same uncomfortable question: how do you certify that a graduate actually knows what their transcript says they know, when a chatbot can draft an essay in seconds?
The answer, increasingly, is not to ban the technology but to rebuild assessment around it. That is the direction the University of Tasmania appears to be heading, with a greater emphasis on supervised and verifiable tasks, closer scrutiny of where and how student work is produced, and assessment design that is harder to outsource to a large language model. It is a pragmatic shift, and one that reflects a broader reckoning in higher education about what a university assessment is actually meant to measure.
Why this is happening now
When tools such as ChatGPT arrived in classrooms, the first institutional reflex across the sector was detection. Universities leaned on text-matching software and AI classifiers to catch students who had leaned too hard on a machine. That approach has aged badly. Detection tools have proven unreliable, prone to both false positives and easy evasion, and they put academics in the invidious position of accusing students on the basis of probabilistic scores they cannot fully explain. The result has been a quiet pivot away from policing individual submissions and towards redesigning the assessment itself.
That is the logic behind strengthening oversight rather than simply adding another layer of detection software. If a written assignment completed at home can no longer be trusted to reflect a student’s own capability, then universities have two broad levers to pull. They can return to supervised settings, such as invigilated exams, oral defences and in-class tasks where a student demonstrates understanding in real time, or they can design assessments that explicitly incorporate AI and then test the higher-order judgement the technology cannot supply. Most institutions, Tasmania included, will end up doing both.
Two ways of seeing it
For university leaders and integrity staff, tighter oversight is a defensive necessity. A degree is a signal to employers and to the public, and that signal collapses if nobody believes the grades behind it. From this vantage point, reasserting control over how assessment happens is less about distrust of students and more about protecting the value of the qualification for everyone who holds one. There is also a regulatory dimension. The sector’s watchdog has made clear that maintaining the integrity of assessment is a baseline expectation, not an optional extra.
Many academics and students see a more complicated picture. A heavy return to supervised exams risks rewarding memorisation and favouring students who perform well under timed pressure, while disadvantaging those with disabilities, caring responsibilities or anxiety. Casual and sessional staff, who mark a large share of undergraduate work, would carry much of the extra burden of redesigning and supervising assessment without a matching increase in pay or time. And students themselves have pointed out the mixed messages: employers now expect graduates to be fluent with AI tools, yet universities often treat the same tools as contraband. The challenge is to teach responsible use rather than drive it underground.
The national picture
Tasmania is not acting in isolation. The Tertiary Education Quality and Standards Agency, the national regulator, has been pressing universities to overhaul assessment in response to generative AI, warning that integrity can no longer be assumed from unsupervised take-home tasks. Its guidance has nudged the whole sector towards what some have called a “programmatic” approach, where a student’s competence is verified across multiple points in a course rather than resting on any single assignment that could be quietly delegated to software.
Group of Eight universities and smaller regional institutions alike have published revised academic integrity policies over the past two years, and the common thread is a move away from the question “did a student use AI?” towards “can this student demonstrate the learning, with or without it?” Tasmania’s announcement fits squarely within that trend, and its significance lies less in novelty than in confirmation that even institutions outside the largest capital-city campuses are now formally restructuring how they assess.
What it means for Australia
The stakes here are national and they are economic. International education is one of Australia’s largest export earners, and its value rests almost entirely on the reputation of Australian qualifications abroad. If offshore employers or governments begin to doubt that an Australian degree reflects genuine capability, the damage would be felt well beyond any single campus. Tasmania, which relies heavily on international enrolments to sustain its regional operations, has a particularly direct interest in protecting that reputation.
There is also a domestic workforce question. Australian employers are rapidly adopting AI, and they want graduates who can use these tools critically rather than blindly. A credentialing system that simply tries to keep AI out of the classroom would produce graduates poorly matched to the jobs waiting for them. The better outcome, and the one the sector appears to be edging towards, is assessment that assumes AI is in the room and tests whether a student can direct it, challenge it and know when it is wrong. That is a harder thing to design and a harder thing to mark, but it is closer to what modern work actually demands.
What’s next
The practical detail will matter more than the announcement. Universities that handle this well will phase changes in with proper support for staff, clear communication to students, and accommodations for those who struggle in supervised settings. Those that simply reinstate traditional exams under pressure risk a backlash and a return to the same credibility problem in a different form. Expect continued guidance from TEQSA, more published policy from individual institutions, and a growing body of evidence over the next year about which assessment models genuinely hold up.
For the University of Tasmania, the immediate task is translating a commitment to stronger oversight into concrete changes that academics can implement and students can trust. For the rest of the sector, it is one more sign that the quiet experiment of the past three years is hardening into permanent change in how Australia decides that a student has learned what their degree claims.
Sources: ABC News Australia.
Comments 1