AI in universities: the case against

This article is part of our Prof v. Prof summer debate series, where academics take opposing viewpoints on hot topics in higher education. Read all the articles from the series here. 

Generative artificial intelligence based on large language models (LLMs) has become a hot topic of debates in higher education. University administrators, professors, staff and students are now asking practical questions about the dos and don’ts, cans and can’ts, and consequences of AI. What can AI do and not do for teaching and research? What AI platforms are best for doing this or that task? How to ensure data privacy and copyrights? How can we prevent academic dishonesty? 

A more fundamental question seems to have escaped the spotlight: Should AI be incorporated into universities’ teaching, research and curricula at all? To answer this question, one must first recognize the moral foundation of utilizing AI in higher education. But it does not have one. People in universities use AI for only one purpose that has nothing to do with ethics: to increase efficiency by making us do more things more quickly, cheaply and easily, which is the core agenda of private businesses. 

Let us take a step back to reflect on what it means to emphasize efficiency in the two pillars of academia: teaching and research. 

Consider the example of assignments. It would be a lot quicker and more convenient for students to complete their assignments and for professors to grade them with the assistance of generative AI. But we do assignments for a reason. Completing assignments helps students master the knowledge and skills they acquire in class through practicing. Assignments allow professors to evaluate the strengths and weaknesses of each student so the former can facilitate the latter’s learning process. 

The entire point of assignments is to compel students to learn outside of the weekly three-hour lecture via a deeper engagement in the materials, as well as continuous conversations with professors. Both students and professors doing their part is the very precondition for an assignment to fulfill its function. Allowing generative AI to take over any component of this process is not merely cheating. It contradicts the very purpose of assignments, and hence of learning and education. 

This is true even in courses where technologies have long been part of the curriculum. A statistics course is one such example. Most statistical curricula require students to use software with some elements of computer programming to analyze data. A statistics course usually has a lab session demonstrating the procedure of using data-analysis software. 

Using software is not just a matter of speeding up calculations. The labs are meant to equip students with practical knowledge about what methodological choices need to be made in different real-world scenarios, and how to use and even customize various functions in a software program according to those choices. Unfortunately, more and more instructors are now simply permitting students to get the syntax for carrying out certain statistical tasks from ChatGPT or Google AI, without even comprehending what it means and how it works. 

Consider the example of literature review as well. In almost all scientific and social scientific research, scholars need to conduct and write up a literature review to summarize what has been said and done regarding their research topics. Students and professors are now increasingly relying on generative AI to do so. After all, LLMs are precisely trained for identifying and replicating patterns of human speeches, writings and even drawings from a large corpus. A well-trained AI chatbot can locate tens of thousands of academic articles and books from multiple databases and generate logical and grammatically correct summaries in a matter of minutes, if not seconds. 

But a literature review is not merely a summary. Research is, after all, an attempt to advance knowledge. To do that, we must first discern the problems, weaknesses and gaps in the existing literature so as to address them and make our own contribution. The originality of our contribution rests heavily upon the originality of our analysis and diagnosis of literature gaps, and hence our own grasp of existing knowledge. 

The analytical nature of a literature review means that once a student or professor lets AI collect and summarize the literature for them, the originality of the review, which is the very foundation of a research project relevant to existing knowledge, is lost immediately. In this sense, asking technical questions about things like the extent of AI chatbots “hallucinating” fake citations is irrelevant. Whether AI expedites a literature review, makes things up or enables academic dishonesty is secondary to the fact that it simply defeats the whole purpose of conducting a review. 

Underneath an entire array of traditions, institutional structures and ways of doing things in academia lies a set of social values that are beyond and often unrelated to efficiency. When we skip the moral question of whether using AI is acceptable, and jump right into the practical ones, we simultaneously stop asking questions about the social values and meanings embedded in the established practices of teaching and research in order to make way for AI. Hence, we let the agenda of efficiency override all other values and purposes that are more important in education. 

The furor over AI in academia is but part of a larger problem. Universities are now offering students general education courses on the know-how of generative AI, such as the University of Alberta’s Artificial Intelligence Everywhere Certificate program. This is, of course, a response to the omnipresence of generative AI in today’s job market. This being the case, the infiltration of AI into academia is just the latest symptom indicating the chronic disease of modern higher education: universities are seen as nothing more than an arsenal for private businesses and governments to produce the next generation of technologies and human capitals. If anything, the obsession with AI in academia reveals that not only the interests but also the mindset of the private sector have parasitized universities. What we need to think about is not how to make use of AI, but the social values and moral basis that higher education will lose after being implanted with a business brain.