Few sectors have felt the impact of Gen-AI as quickly and directly as HE education. Given the speed and scale of this change, it is unsurprising that much of the institutional debate so far has focused on whether students will use AI to cheat, whether existing assessment practices remain fit for purpose, what AI means for creativity and critical thinking, and whether assessment can, or should, be made AI-proof.
While these questions are important, they frame Gen-AI mainly as a challenge of student behavior and assessment design. Drawing on our recent research with co-authors, we urge educators, students and university leaders to consider a deeper question: why and how is Gen-AI reshaping the very ownership and production of knowledge in higher education? Asking that question matters, for Gen-AI is unlike other technologies; a telescope helps us see distant stars, but Gen-AI outputs have world-making implications. Their ‘synthetic’ outputs are already now profoundly shaping research and teaching in HE.
Gen-AI and the Governance of Knowledge
It follows that it is important to recognize that Gen-AI is not simply a ‘tool’ that support users in completing tasks. It shapes the conditions under which knowledge is created, interpreted and legitimized. In so doing, itinfluences which sources are used, how ideas are summarized, how arguments are framed, and what comes to appear credible or reliable knowledge.
For this reason, the use of Gen-AI in HE cannot be treated as a technical or pedagogical matter alone. Its ‘inner’ workings (combing words through probabilities rather than genuine understanding of meaning) create ‘outer’ consequences (a social world shaped through probabilities rather than genuine human understanding). Therefore, Gen-AI must also be understood as an epistemic and institutional issue, because it affects not only how academic work is done, but how knowledge itself is produced, circulated, and applied eventually.
This is especially important when leading universities such as the University of Oxford, and UNSW Sydney among others, enter into institutional agreements to provide staff and students with access to ChatGPT Edu. These agreements do far more than just widen access to Gen-A: they move it from individual choice into institutional practice and expectation, giving its use formal legitimacy across research and teaching in HE. These developments cannot be celebrated as institutional innovation, for there is in fact little to celebrate when Gen-AI becomes embedded in the knowledge-generation work in universities.
From Knowledge Producers to Knowledge Consumers
The risk is not that Gen-AI will occasionally produce poor essays or inaccurate answers. It is that regular reliance on these systems will normalize a more passive relationship with knowledge. When students and academics use Gen-AI to frame arguments, synthesize literature or generate conclusions, they become less involved in the difficult work of thinking through complexity themselves. That is, asking fundamental questions, such as knowing-why and knowing-how, and how these questions tie learners to the time and space they inhabit, is an endeavour increasingly less pursued by students and academics alike.
Over time, this will inevitably can change the human role in higher education, for the worse. The danger is that polished AI-generated text can look like knowledge, even when the difficult work of judging, questioning and contextualizing ideas has been bypassed. So, instead of actively making sense of ideas, students and academics will likely find themselves simply polishing, checking or repackaging outputs generated by systems whose underlying logic they cannot fully see. At that point, they will have transitioned from knowledge producers to knowledge verifiers.
The Problem of Organized Immaturity
In our recent research, we relate this development to the concept of “organized immaturity”, which implies a condition in which social and technological systems gradually weaken people’s capacity to use their own reason in public, critical and independent ways.
In HE, this can happen in three connected ways. First, Gen-AI can encourage infantilization, where people become more willing to defer reasoning to automated systems. Second, it can encourage reductionism, where ambiguity, judgment and creativity are flattened into statistical patterns. Third, it can encourage totalization, where a technology becomes so embedded in everyday work that it becomes difficult to imagine teaching, learning or research without it.
We are concerned that reliance becomes routine, then expected, and eventually hard to question. The question is less whether Gen-AI is thought to replace human judgment entirely; it suffices when it makes it easier for humans to simply rely on Gen-AI outputs first. The erosion of judgement unfolds over time then.
The University’s Responsibility for Knowledge
If HE is to avoid organized immaturity, it must protect the conditions that make independent thought possible. That means asking who owns the infrastructure through which knowledge is produced and circulated?What forms of thinking are encouraged or discouraged? And how academic freedom, disciplinary diversity and intellectual independence can be preserved?
The future should not be one in which academics and students mainly verify, edit and repackage machine-generated outputs. Nor should it be one in which universities become dependent on systems whose assumptions, priorities and limits are not transparent and cannot be easily challenged.
Given Sam Altman’s future vision of AI where it becomes a ‘utility’ (like water) that we buy from OpenAI ‘on a meter’, universities and academics alike must confront what Gen-AI is doing to the ownership, production and governance of knowledge before the capacities that make education matter (e.g., independent thought, critical judgment, democratic responsibility and the freedom to ask difficult questions) are irreversibly lost.
