Daily Maverick
Dailymaverick logo

Opinionistas

This is an opinion piece. The views expressed are not that of Daily Maverick.....

ChatGPT poses an existential threat to universities and the integrity of degrees

Andy Carolin
By Andy Carolin
2 May 2025
Facebook
2
Advanced critical, independent and strategic thinking is the basis of the future of work, and universities owe it to their students to ensure they graduate with the very skills and knowledge that their certificates purport to show.

Those of us who work in universities understand the rich, engaging and transformative benefits of the well-conceptualised courses and qualifications that universities offer.

University classrooms are places for dialogic and collaborative learning, engaged citizenship, conceptual contestation and the facilitation of advanced critical thinking skills.

But for the hundreds of thousands of prospective students who desperately pursue admission to a South African university each year, the motivation is simpler: to get a qualification that will secure employment and the possibility of social mobility.

So, what’s the problem?


The impact of a degree or diploma certificate on employability hinges almost entirely on the extent to which employers recognise the value of that qualification. Within the labour market, then, universities play a key certification function, where a qualification certifies that a core set of skills and knowledge has been acquired. (Of course, there are also statutory bodies that oversee the qualifications in a more formal sense).

But we are now more than two years into the easy availability of generative AI such as ChatGPT. This means that a large part of the current cohort of students has had access to a tool that allows them to submit almost-always passable written submissions.

This is in a context where many universities have not returned to traditional in-person invigilated exams after the massification of online assessments during the Covid-19 lockdowns.

To put it simply: I have no doubt that there are students who used ChatGPT consistently over the past year or two to cheat their way through degrees and diplomas without formulating original ideas, applying important concepts, or even reading prescribed material.

This means that we risk an ever-increasing number of students who hold certificates that fraudulently certify their mastery of skills and content knowledge that some of them may have only barely attempted.  

But surely it’s just progress?


I should point out that many of my academic colleagues have evaluated developments in AI and have come to different conclusions. For many of them, the benefits and inevitability of AI outweigh the perceived risks. The arguments that have been made include:

  1. “Let’s rethink what we are assessing”: The introduction of AI has fired up some necessary conversations around probing what is assessed, how it is assessed, and why. These are important questions, especially given the urgency of developing a wide range of skills for students to be able to participate in an unknown future workforce. But creating opportunities for collaboration, creativity or communication, for example, does not always neutralise the risks posed by ChatGPT’s superlative ability to mimic insightful descriptive text;

  2. “Luddites”: Those of us who are deeply worried about generative AI are often dismissed as being unreasonably resistant to change, described by one scholar as “Luddites” and by another as “vilifying innovation”. It is often argued that concerns over generative AI are no different to concerns over the impact of the internet two decades ago. But I think this oversimplifies the case: while plagiarised content that is available on the internet has a clear linguistic and substantive fingerprint that can be traced back to its original source, the same cannot be said for the almost infinitely “original” content generated by AI. Put differently, when it comes to it, students can simply deny using generative AI and there is simply no surefire way to prove otherwise;

  3. “To compete in the future, all students must be able to work with AI”: This, of course, is undoubtedly true. In fact, I believe that most degrees should now include compulsory modules on AI. I have seen first-hand the extraordinary interventions being developed to prepare students for a digital future (see, for example, here and here). But just because AI literacy is important, it does not mean that it should be casually allowed to overrun all other learning areas. Being able to discern the accuracy of AI-generated text, to use advanced technologies, or to formulate appropriate AI writing prompts, for example, does not replace the depth of knowledge and critical independent thinking that traditional degree programmes offer. AI literacy skills should be developed, and modules should teach and assess core disciplinary knowledge and critical thinking skills. But there is no reason that the use of the former should be allowed to replace the latter; and

  4. “If we trust students, they will be ethical”: While this argument is often advanced by well-meaning academics who believe in inculcating cultures of integrity, it tends to rely on the idea that academic staff can manifest ethical conduct by sheer force of will. But the evidence globally is overwhelming that many students will cheat if they are unlikely to face consequences (see here, here and here).


‘But where are we supposed to get our ideas?’


A group of my undergraduate students recently responded with confusion when they were told that they may not use ChatGPT in their upcoming assessment, with one student asking: “But where are we supposed to get our ideas?”

The other students nodded vigorously, looking bewildered. This question shocked me: universities serve a key social and economic function in which graduates are equipped with advanced critical thinking and problem-solving skills.

It is these skills that are developed across university curriculums, ranging from disciplines such as literary studies, art history and philosophy, to modules in professional programmes such as contract law, educational psychology and public administration.

But with some students relying on ChatGPT for their assessments, they do not need to learn how to formulate original ideas or respond independently and innovatively to problems, nor do they ever have to master skills such as synthesising contrasting perspectives or organising their ideas logically.  

So, what can we do?


Academics across many institutions were recently left dismayed by a Daily Maverick article by Rebecca Davis which quoted university leaders who seemed to be underestimating the challenges of AI (“CheatGPT crisis – SA universities faced with a burgeoning degree of AI-written academic assignments”, 5 April 2025).

The responses from various institutions appeared disconnected from our experiences on the frontline. While developing alternative assessments is a valuable intervention that we must pursue with rigour, we also need to recognise that fostering certain skills, content knowledge and intellectual orientations might require more traditional assessment models.

Many bold academics are starting to insist on a return to in-person discussions, tests and examinations that assess the application of core knowledge and critical thinking skills.

This may go against the enthusiastic embrace of technological innovation, but, for many disciplines, I am left wondering: what other choice do we have? DM

Categories:

Tags: