There is a comfortable story that Australian universities like to tell themselves about artificial intelligence, and it goes something like this: the technology is moving faster than any institution can reasonably keep pace with, the tools are immature, the risks to academic integrity are real, and so a measured, cautious posture is not timidity but prudence. It is a tidy narrative. It is also, according to one senior figure at the University of Sydney, largely a way of avoiding a harder conversation.
Speaking on the higher education podcast HEDx, the University of Sydney’s Lucy Marshall put the point bluntly, as reported by Campus Review: universities do not have an innovation problem when it comes to AI. They have a culture problem. The distinction sounds academic until you sit with it, and then it starts to sting. If the barrier were a lack of ideas, more grants and more pilots would fix it. If the barrier is culture, the fix runs straight through how institutions reward, govern and trust the people inside them.
What Marshall is actually naming
The argument, as it came through on the podcast, is that the sector is not short of clever people building clever things. Anyone who has walked a corridor at a large Australian university knows the place is thick with pilots, working groups, learning-and-teaching experiments and enthusiastic early adopters quietly wiring generative tools into their marking, their tutorials and their research workflows. The energy is there. What is missing, on this reading, is the connective tissue that lets any of it scale: a shared willingness to take a risk, to be seen to fail, to move a promising experiment out of one enthusiastic department and into the institution’s bloodstream.
Marshall’s framing lands on something most leaders privately recognise. Innovation gets celebrated in the abstract and starved in the particular. A lecturer who redesigns an assessment around AI use rather than banning it is doing exactly what the strategy documents ask for, and yet often finds themselves navigating suspicion, compliance reviews and colleagues who regard the whole exercise as cheating with extra steps. The problem is not that the idea does not exist. The problem is that the culture makes the idea expensive to hold.
The other side of the argument
It would be too neat to treat culture as simply the villain here, and the caution has defenders worth taking seriously. Universities are not startups, and they are not meant to be. They certify learning, and that certification is only worth something because it is trusted. When a cohort graduates, employers and the public assume the degree means the graduate can actually do the thing. AI tools that can draft an essay, solve a problem set or summarise a body of research in seconds put genuine pressure on that assumption, and the people responsible for academic standards are right to be nervous about moving fast and breaking the one thing they cannot afford to break.
There is also a labour dimension that a culture critique can gloss over. Academic and professional staff across the sector are stretched, and many have watched wave after wave of technology arrive with promises of efficiency that translated, in practice, into more work and more surveillance. A staff member who greets the latest AI mandate with folded arms is not necessarily a cultural laggard. They may simply have learned, from experience, that the person who volunteers to trial the new system is the person who inherits the mess when it misfires. Culture, in other words, is downstream of incentives, and the incentives in a modern university do not obviously reward the risk-taker.
Both of these things can be true at once. Caution can be legitimate and culture can still be the binding constraint. Marshall’s point is not that universities should abandon their standards, but that they keep reaching for the language of innovation when the actual work in front of them is the slower, less glamorous business of changing how the institution behaves.
Why this matters for Australia
The stakes here are not confined to one Sydney campus or one podcast. Australian higher education is a genuinely large national enterprise, one of the country’s biggest export earners, and a sector the federal government is leaning on to help drive the productivity story it keeps promising. The Office of AI and the various national frameworks now taking shape all assume that universities will be both a supplier of AI-literate graduates and a proving ground for how the technology gets used responsibly. If the sector’s real bottleneck is cultural rather than technical, then more funding programs and more strategy refreshes will not move the needle much.
There is a competitive edge to this too. Employers are already reshaping entry-level work around AI, and students can feel the ground shifting under the degrees they are paying for. A university that treats AI primarily as a threat to be policed risks turning out graduates who learned to hide their tool use rather than to reason with it, which is close to the opposite of what the labour market now wants. The institutions that get the culture right, that make it safe to experiment in the open and to be honest about what works, stand to become far more attractive to students, staff and industry partners than those still litigating whether the technology should be allowed on campus at all.
Australia also has a specific vulnerability. The sector’s heavy reliance on international enrolments means reputation travels fast and cuts both ways. A national higher education system that is seen as slow, defensive and internally conflicted about AI does not just lose a few pilots. It risks ceding ground to competitor systems in the United Kingdom, Canada and Asia that are further along in normalising the technology inside teaching and research.
What comes next
The uncomfortable thing about a culture diagnosis is that it does not come with a purchase order. You cannot buy your way out of it the way you can buy a new learning-management system. Changing a university’s culture means changing what gets rewarded at promotion time, how failure is treated, and whether the people experimenting at the edges are backed or quietly managed out. That is leadership work, and it is far harder than announcing another AI taskforce.
Marshall’s intervention is valuable precisely because it refuses the easy version of the story. The sector will keep producing innovation, because it always has. The open question, and the one worth watching across Australian campuses over the next couple of years, is whether the institutions can build a culture willing to actually use it. On the evidence so far, that remains the genuinely unsolved problem.
Sources: Campus Review.
