The quiet assumption that generative artificial intelligence was a fringe habit among a few enterprising students has collapsed. According to new analysis published in The Conversation, more than half of Australian university assignments now involve some use of AI, a figure that shifts the conversation from whether students are using these tools to what universities intend to do about it.
That threshold matters because it reframes the problem. When a technology touches a minority of submissions, institutions can treat it as a discipline issue, something to be policed at the margins. Once it crosses into the majority, policing stops being a credible strategy. The behaviour has become normal, and normal behaviour cannot be stamped out with warnings and detection software. It has to be designed for.
How the sector arrived here
The trajectory has been steep. ChatGPT arrived at the end of 2022 and within a single semester Australian universities were scrambling to write policy for a tool that most academics had not yet touched. The first instinct across much of the sector was defensive. Some institutions leaned on AI-detection services, others returned to handwritten exams and invigilated assessment, and a few issued blanket prohibitions that were quietly unenforceable from the moment they were written.
None of those responses held. Detection tools proved unreliable, flagging honest work as machine-generated and missing genuine misuse, a failure that carries real consequences when a student’s academic record is at stake. Blanket bans ran into the awkward reality that the same tools students were told to avoid were being embraced by the employers those students would soon join. The gap between campus rules and workplace practice widened by the month.
What the latest finding confirms is that students did not wait for the sector to make up its mind. They adopted the technology at the pace of consumers, not institutions, and they folded it into the ordinary rhythm of essay drafting, coding, summarising and revision. For a large share of assignments, AI is now simply part of how the work gets done.
Two ways to read the number
There are two broad camps in how the sector interprets this, and they lead to very different responses.
The first sees a crisis of integrity. On this view, if a majority of assignments involve AI, then the value of the degree itself is in question. Employers hire graduates on the assumption that a qualification certifies capability, and if that capability can be outsourced to a chatbot, the certificate loses meaning. Advocates of this position argue for a return to supervised, in-person assessment, where what a student can do is observed directly rather than inferred from work produced in unmonitored conditions. It is a demanding and expensive path, but it protects the one thing a university ultimately sells, which is a trustworthy signal of what a person knows.
The second camp sees an overdue reckoning with how assessment was designed in the first place. If a language model can complete an assignment to a passing standard, the argument runs, then the assignment was probably testing recall and formatting rather than judgement, and it was a weak measure of learning long before AI existed. From this angle the task is not to defend the old assessment but to build a better one, setting problems that require students to apply reasoning, defend choices, and demonstrate the kind of thinking that a general-purpose model cannot fake on its own. Supporters of this view often want AI use made explicit and taught, on the basis that graduates who cannot use these tools well will be at a disadvantage in the workforce.
Most working academics sit somewhere between the two, and the practical reality is that universities will need elements of both. Some assessment has to be secured so that a genuine baseline of individual capability is verified. The rest has to be redesigned so that AI use is either irrelevant to the outcome or an intended part of it.
What it means for Australia
The stakes here are unusually high for Australia because international education is one of the country’s largest export earners, sitting behind iron ore and coal in economic weight. That entire industry rests on the credibility of the qualification. Overseas students, their families and the governments that recognise Australian degrees are paying for a signal of quality, and if that signal is seen to be corroded by unchecked AI use, the reputational damage would flow straight into a sector already under pressure from visa settings and cost-of-living strain.
There is a domestic workforce dimension as well. Australia has spent years talking about a productivity problem and a skills shortage, and the graduates moving through universities now will spend their careers working alongside AI systems. A higher education system that teaches students to use these tools with judgement, to check their output and understand their limits, produces a more capable workforce than one that pretends the tools do not exist. The risk is a two-tier outcome, where some universities invest in redesigning teaching while others fall back on bans they cannot enforce, leaving their graduates less prepared.
The regulator has a role to play too. The Tertiary Education Quality and Standards Agency has been pushing institutions to take a principles-based approach to AI rather than a purely punitive one, and the pressure on universities to show they have credible, defensible assessment practices is only going to grow as the numbers keep climbing.
What comes next
The near-term response is likely to be uneven. Expect more programmatic assessment, where a student’s capability is judged across a whole course rather than a single essay, alongside a revival of oral examinations and in-class tasks for the components that genuinely need to be secured. Expect clearer disclosure rules, asking students to declare how and where they used AI, on the understanding that transparency is easier to manage than prohibition.
The harder work is cultural. Academics need time and support to redesign assessment, and that is a resourcing question at a moment when many institutions are cutting costs. Whether universities treat this finding as a threat to be contained or a prompt to modernise will shape the credibility of Australian degrees for the next decade, and the majority has already made its choice.
Sources: The Conversation.
