On the one hand, there is no group for whom text-generating Large Language Models are less suitable than university students. On the other hand, there is no group for whom text-generating Large Language Models are more suitable than university students.
On the one hand, at this optional third stage of education, students are undertaking increasingly sophisticated and independent work. Less and less of what they learn is rudimentary, and less and less of what they do with that knowledge is mere recitation. Students are at university to gain a deep understanding of complex, organized systems of thought, and to make some creative contribution to those systems. The essay and dissertation are the ordinary forms of that contribution. It is not, in general, expected that a student’s writing will be very significant for the relevant discipline, but it is meant to be very significant for the student herself: she will have thought hard about something complicated, and exerted herself to put her thoughts into coherent prose. It is because she has undertaken such work that her studies are considered valuable both for her fulfilment as a rational being, and for her future social prospects. But this value is reduced to nil if the student has not in fact undertaken the work, but merely entered prompts into a software program, in order to produce a simulacrum of it. Nor would any student who herself wished to learn what her studies are meant to teach have any desire to do such a thing.
It is not expected that a student’s writing will be very significant for the relevant discipline, but it is meant to be very significant for the student herself.
On the other hand, the mere detention of a university degree, regardless of what work has gone into obtaining it, can serve as a necessary token for entry into certain jobs and for overall social success. Students attend university to attain this token, often without the interests or abilities needed to write the essays and dissertation they must perform to get it. Happily a vast amount of data exists to enable LLMs to fill this gap. Nor is the main weakness off LLMs, their “hallucinations”, necessarily an obstacle to their usefulness for the student: the work she is doing is after all a mere exercise, with no consequences in any world beyond the university’s sealed environment. If the essay contains unreal claims and references, there are no dangerous consequences, so long as the student and her teachers don’t mind. The student need not mind, as long as she gets her token at the end of the process. Her teachers have good reasons not to mind: their jobs are dependent on students getting through their degrees, while the policing of LLM use is sickeningly tedious and riddled with uncertainty.
Of the two above scenarios, one could easily state that the first represents a deluded dream, and the second the sordid reality. Yet that would be overly cynical. Both the perfectly diligent and knowledge-loving student, and the utterly incompetent and lazy one are ideal types. Some students do almost match these types, and many more are strongly similar to one or the other. Universities really do have some extremely serious and eager students, and they also really do indulge students who would not be there if they didn’t see a degree as a passport, and don’t care much about how they obtain it. Yet most students contain strong elements of both types within them: like the rest of us, they are both tempted by easy shortcuts and drawn to the nobility of strenuous exertion.
Both the perfectly diligent and knowledge-loving student, and the utterly incompetent and lazy one are ideal types
It has always been thus. Since their medieval beginnings, universities have never existed for some purely contemplative pursuit of knowledge or wisdom: part of what they offer has always been a path to worldly advancement. That is not bad in itself: there is no shame in wishing to make one’s way in the world, and literacy-based knowledge isn’t intrinsically debased by being of practical use. But the instrumental aspect of university studies means that they contain an in-built incentive to cheat. This could only be eliminated if the obtention of a degree came with no advantages beyond the intrinsic value of whatever one had learnt and done in one’s studies. Such a state of affairs is scarcely possible in itself, and approaching it would collapse universities.
Thus the curse of LLMs for universities is not that they have tainted what was once pristine. Rather, it is that they are a perfect tool for the utter debasement of all that is already corrupt and corruptible within students, lecturers and the institutions that encompass them. Accordingly, unless and until LMMs become prohibitively expensive, they will not be driven out of universities, no matter how much students are policed or their assessments are re-formatted.
LLMs could only be defeated by a deep belief, shared between students, lectures and administrators, in the great worth of thinking and writing without these machines. On the one hand, universities are quite incapable of fostering such a belief to the exclusion of opportunism and abuse. Yet, on the other hand, universities are still the only sphere, among the great institutions of modernity, where such a belief can be easily understood at all.
