Since the launch of ChatGPT in 2022, text-based professions have been wondering how these large language models will impact their work. Academia and education have been particularly troubled with questions about what the generalisation of these models in web browsers and what ‘ai-enhanced’ social media will do to young people’s capacity to experience, feel, learn and write about this world. Now that we academics see more and more of those essays that seem too well written to be true, we actually do have to reckon with the difficult question what to do with these genAI bots that treat ‘knowledge’ like instant alphabet soups in grammatically attractive packaging. Every colleague in social sciences and humanities has an anecdote to tell how they were initially bluffed by a genAI essay they had asked a LLM to write, thinking that they would get googlish gargle. Yet, those models actually propose (in an affable tone of overly gentle chattiness) sleekly written and, on the surface, sensibly argued essays! It is only at a much closer look that one realizes that this style is not elegant but bland; that the sentences are not clever insights but mere descriptions dressing up as important thoughts; that paragraphs are rows of twaddle statements without inner relations of reasoning; that the entire essay is an exercise of filling paper with words.
For many colleagues, especially in those disciplines that rely on essay as assessment of learning progress, it has become, depending on one’s nature, a sport or a chore to find ways to beat the students at the game of generating AI essays. Like in the olden times of plagiarism when Google and Wikipedia already existed but TurnItIn didn’t, we find ourselves frantically checking quotes, citations and references to find hallucinations. We generate AI essays and do linguistic comparisons of the students submissions’ grammar, vocabulary and essay structure in order to be able to reach some estimation whether they have used genAI wholly, mostly, partially or a wee bit in their essay. We cite students to our offices to test them on their writing. We curse through handwritten essays and notes. And we treat every well formulated sentence, every clever thought, every smart insight or brilliant show of knowledge as utterly suspicious. We do not assess their learning but their capacity, willingness and ability to cheat. No better way to poison the relationship between teachers and students.
It is thus unsurprising that the problem of cheating students has received the most attention in the discussion of the future role of AI in universities. Declaring defeat in wanting to ban genAI, most universities have in the past year tried to walk a thin line between advising students on smart uses of AI and hoping for the arrival of reliable detection software (which in my guess will come as soon as one of the AI companies will have secured a sufficiently large share of the education market to lock institutions into buying their specifically designed detection software). Some colleagues have even resorted themselves to using genAI, for instance to develop chatbots that, service centre like, should answer common student questions on class materials. We might soon find ourselves in the absurd situation that academics and teachers will use genAI bots to create questions which students will ask genAI bots to answer.
While this focus on cheating is not unjustified (after all, doing really nothing, not even handwriting a cheat sheet in tiniest letters, for a mark is really, really unfair), it is nevertheless misleading. The problem university education is faced with by LLM is not student dishonesty. It is that it is inescapably confronted with one of the most profound contradictions of its self-justificatory discourses: claiming to be spaces of excellent thinking, innovative creativity and outstanding intellectual originality, universities are, by and large, refining processors of conventional knowledge. Despite all talk of ‘out-of-the-box’ reasoning, of ‘critical thinking’, of ‘excellence’, ‘innovation’ and so on, the large majority of university teaching is dedicated to bringing the greatest number of students over the greatest number of conventionally determined certification hurdles to slate them into their categories of ‘merit’ in a well stratified job market.
Whether we are talking about being trained in biogenetic engineering, writing marketing pitches or legal opinions on neighourhood fencing disputes, undergraduate students in our universities learn not to become future Nobel prize winners but to fill white collar jobs which are nothing more than any one of the zillions of tiny screws that keep our capitalist systems working. The most egregious of those have been described by the anthropologist David Graeber as ‘shit jobs’ and it is many of those that are currently also, unsurprisingly, those in danger to be replaced by genAI. These are jobs where there is no originality or critical thinking needed. On the contrary any of those intellectual qualities that allow critical thinking would be even counterproductive. Imagine that biolab technician, that marketing person or that lawyer who is all of the sudden asking fundamental, critical, out-of-the-box questions about the ethics, social usefulness or fairness of what they are doing…chances are that they will not be in their job for a very long time…

In practice, too, our education systems are not organized in a way that this kind of critical thinking allows that kind of originality which will lead to alternative views of this world. Under the pressure to serve ever more students with ever less staff, the university world (which is not only academics but also administrators and regulators in and outside universities) has resorted to more and more standardization in the name of ‘quality control’.
What is striking if a medieval university class is compared with a modern one is not that there have always been students chatting, not paying attention, sleeping, eating or being lazy in class, but how tiny those classes were. Education has up to modern times been the privilege of a incy wincy tine elites. Plato whose resistance to the written word is often cited in the current debates as a parable for resistance to new technologies taught to a class of a handful students; and that apparently over entire days where teacher and students would not only think together but also eat together, have strolls through gardens and leisure to break for a moment before gathering again.
Modern universities, too, have been up to today institutions of elite distinction. What was particularly obvious in pre-modern universities, in which sons and daughters of the nobility were learning (yes, daughters, too, it was only in modern times that women were banned from universities), has, however become less obvious in today’s higher education systems. Some university systems still cultivate institutionally the social distinctions between those they are training for literate professions that modern capitalism needs, and the elitist selectivity of ‘arts and sciences’; think for example of the British distinction between old and new universities (Oxbridge v Loughborough, Leicester or Lancaster…) or the German distinction between ‘professional’ and academic universities (Fachhochschule v Universität). Yet, more importantly, with the modern massification of literacy and education, including university education, the distinction processes that separate the elitist wheat from the common shaff has been displaced into the classroom.
This is less the result of the massification of higher education as such, but of the financialised, neoliberal management reaction to it. With the shift of financial and decision power from academics to managers in today’s universities (expressed among others by the fact that academics now count mainly as ‘expenses’ in the university budget), managers have consistently sought to make universities ‘productive’ by squeezing ever more students into ever more expensive classes with fewer and fewer resources. Classes have become bigger, teaching hours fewer and complex knowledge more and more standardised and simplified. As a result, the only students who actually can go beyond simply absorbing the minimum conventional knowledge are those who have resources to study outside the university. That are commonly thoose who do not spend hours working to pay their rent; those who have grown up surrounded with books and the certainty that the adult world and those books potentially contain an answer to any question the curious child might have had; those who are supported by a family and amical environment that appreciates intelligence, wit, curiosity and eloquence; those who are cared for by a family and who have not care for them. It is rarely those who have the intellectual and power will resources to fight their way up through adverse university systems. Universities are, in the words of Sarah O’Shea, Olivia Groves and Janine Delahunty who have analysed the first-in-the-family graduates’ job chances, ‘far from a level playing field’.
In the perspective of shovelling ever more students through university while continuously reducing the costs of offering an education that would really create a level playing field, the promises of genAI are enormous. Creating real equity would require universities not only to put many more human resources and value into teaching and learning; it would also require universities, politics and larger society to show more recognition and support for students, first and foremost financially. The self-teaching university is the wet dream of university managers and politicians who want to believe that the problems of social organisation that are reflected in the classroom can be solved through technology. Technology companies have excelled in the past to make universities (and other businesses) believe that any of their new inventions whether PowerPoint, the internet, online learning platforms like Moodle, Mass Open Online Classes, Zoom or now genAI would allow them to ‘produce’ teaching content for an ever-greater number of students with ever smaller input. Never mind that there are numerous studies that show the contrary (Emails are a case in point) – when it comes to ignoring research (often from academics in their own uni) that proves their neoliberal worldview wrong, it is especially university managers who prove their qualities as intellectual contortionists.
Except that now with genAI the ‘productivity gain’ is also on the side of the students and that is what comes back to bite.