Generative artificial intelligence has moved from novelty to fixture inside Australian education faster than almost anyone in the sector predicted, and the numbers now make that shift impossible to ignore. According to a discussion hosted by Curtin University, close to 80 per cent of Australian students are already using generative AI tools such as ChatGPT and Microsoft Copilot in their studies. That is not a fringe cohort experimenting at the margins. It is the mainstream, and it lands squarely on institutions that in many cases are still working out where the line between assistance and cheating should sit.
The conversation, part of Curtin’s long-running podcast series, frames the technology as neither saviour nor threat but as something more awkward: a tool that is genuinely useful, widely available, and difficult to police. That framing matters because it moves the debate past the reflexive positions that dominated the first eighteen months after ChatGPT launched, when universities lurched between outright bans and cautious pilot programs. The more interesting question now is not whether students use these tools, but what the near-universal adoption does to the value of a degree, the design of assessment, and the relationship between teacher and student.
How the classroom changed so quickly
The speed of the shift is worth sitting with for a moment. When large language models first reached the public, the standard institutional response was containment. Some schools blocked the tools on their networks, some rewrote academic integrity policies overnight, and a cottage industry of AI-detection software promised to catch offenders. Almost all of those detectors proved unreliable, flagging innocent students and missing sophisticated misuse in roughly equal measure. The Curtin discussion picks up the story from there, at the point where containment has quietly failed and adaptation has become the only realistic path.
What replaces containment is harder work. It means rethinking assessment so that it measures understanding rather than the ability to produce a polished paragraph. It means teaching students how to interrogate an AI-generated answer, spot its confident errors, and cite it honestly. And it means accepting that a generation entering the workforce will be expected to use these tools competently, which makes a blanket ban look less like protecting standards and more like withholding a skill employers now assume graduates have.
Two ways of reading the same data
There are broadly two camps in this argument, and both have a fair point. The optimists see generative AI as a genuine leveller. A student who struggles to structure an essay, a learner working in their second or third language, or someone juggling study with shift work and caring responsibilities can use these tools to draft, revise and understand material that might otherwise have been out of reach. On this reading, the 80 per cent figure is a sign that a useful technology has diffused quickly and fairly, and the job of educators is to channel it rather than resist it.
The sceptics counter that convenience has a cost. If a tool can produce a competent answer in seconds, the risk is that students stop doing the cognitive heavy lifting that learning actually requires. The worry is not plagiarism in the narrow sense but a hollowing out of the struggle that builds real capability, the slow business of wrestling with a problem until it makes sense. That concern has been echoed well beyond Curtin. FluentSea has previously reported on the argument that banning AI in universities will not prepare students for the future, and on separate warnings that the sector has a cultural problem with how it talks about these tools rather than a purely technical one. Both things can be true at once: the technology can widen access and dull effort, depending entirely on how it is used.
What it means for Australia
For Australia specifically, the stakes are unusually high because education is not just a public good here, it is one of the country’s largest export industries. International students underwrite the finances of most major universities, and the credibility of an Australian qualification abroad depends on the perception that it certifies real learning. If assessment fails to keep pace with AI, that credibility erodes, and it erodes quietly before anyone notices the damage. Regulators and vice-chancellors know this, which is why the Tertiary Education Quality and Standards Agency has been pushing institutions to redesign assessment around what it calls trustworthy evidence of learning rather than relying on detection.
The pressure runs into schools as well. Primary and secondary teachers are grappling with the same tools without the research infrastructure that universities can draw on, and state education departments have issued patchy and sometimes contradictory guidance. There is also an equity dimension that rarely gets enough attention. The best generative AI tools increasingly sit behind paid subscriptions, which raises the prospect of a two-tier system where students who can afford the premium models gain an advantage over those who cannot. A technology sold as a leveller could, if left alone, become another line of disadvantage.
What comes next
The honest answer is that nobody has fully solved this, and the Curtin podcast does not pretend otherwise. The near-term trajectory points towards more oral assessment, more supervised in-class work, more assignments that ask students to show their reasoning and their use of AI rather than hide it, and a slow rewriting of integrity policies to reflect a world where the tools are assumed rather than banned. Professional development for teaching staff is the quiet bottleneck here, because policy written in a senate committee means little if the lecturer at the front of the room has not been shown how to apply it.
What seems clear is that the 80 per cent figure will not go backwards. The generation now moving through Australian classrooms will graduate into workplaces where fluency with these systems is a baseline expectation, not a bonus. The institutions that treat this as a design challenge rather than a discipline problem are the ones most likely to come out ahead, and the discussion coming out of Curtin is a useful marker of how far the conversation has already travelled from the panic of a couple of years ago.
Sources: Curtin News
