Back in the day, we had Cliff Notes. These short pamphlets summarized the plots of great literary works and provided additional information (e.g., symbolism, themes) that one might expect to be tested on by, say, a high school or college teacher. As a high school sophomore facing a particularly brutal schedule I once used them instead of reading Charles Dickens’s A Tale of Two Cities. I got an A on the exam and never ended up reading the book.
With the wide availability of AI tools, and particularly LLMs, today, I would not want to invest in Cliff Notes. Everyone now has access to these book summaries, and it is easy enough to generate short answers to the kinds of questions that might appear on an AP exam. From what I can see, these summaries and answers are quite good. I suspect that more than a few students are reading only a fraction of the assigned reading in their literature classes, and still getting good grades.
The use of Cliff Notes and LLMs raise a thorny educational question. The focus of pedagogical anxiety is often on evaluation, but the real problem runs deeper: it is the confusion between process and product.
This problem is actually ancient and was a lively topic of conversation among Medieval philosophers. Revelation, it was commonly thought, was true and relatively complete. God, out of concern for the welfare of human beings, provided a set of true guidelines for living in the world. Jews, Christians, and Muslim all presumed that answers to most of their basic questions could be found in their Scriptures. One doesn’t need a debate and long set of complicated arguments to know that murder or adultery was wrong; it’s written right there!
Yet these same religious thinkers then hit upon the next natural question: What, then, is the point of reason, which also has to be recognized as God-given? And what is the relationship between the products of reason and philosophy and the truth of Scripture? This problem was sharpened by the fact that most religious thinkers in the Middle Ages subscribed to some form of Aristotelian philosophy. Why bother with Aristotle when the truth was right there in Scripture? While some of the participants in this debate simply elevated the role of Scripture over philosophy, most saw value in combining the two.
The answer to this problem took several forms. Briefly, nearly everyone began with the assumption that the truths that emerge from revelation and reason were identical. At the same time, they recognize that human reasoning is a gift of God and therefore must be given to us in order to promote our own flourishing. Yet since reason alone can sometimes lead us astray, God revealed our ultimate destinations; revelation serves a a kind of high-level road map. For those without ability or training in philosophy, this might be enough.
For the great religious thinkers and their students, though, this was entirely insufficient. I cannot become a better person, they argue, unless I go through the difficult task of discovering truth for myself. The process of reasoning is transformative. It awakens me to the real truth in all its complexity. The end proposition (e.g., observe the Sabbath or do not drink alcohol) is necessary but almost beside the point. In more religious terms, only reason can bring one to true knowledge of God.
One of the most notable features of most LLMs I have used is its oracular quality. Even a Google search immediately makes visible the fact that nearly every important question has many different possible answers and sources. Not so for the slick and confident AI bots, which even shift seamlessly when challenged. Even when citing sources, its tone is hardly one of debate. The language of LLMs is that of revelation.
As readers of my Substack and blog know, I am a (skeptical) AI fan. I think that these AI tools will transform the nature of education and that they offer opportunities for enhancing pedagogy. Their integration into college courses, though, is not without danger. That danger is the confusion between process and product.
There are many times in life where the final product matters. An equation or procedure can be a matter of life and death. A PowerPoint presentation can result in an enormous deal. A piece of writing can be accepted or rejected, sometimes with enormous consequences for the author.
But much of humanities (and sometimes non-humanities) education is about the process. The skill of academic paper writing is not, in itself, a useful skill; most people will never write another one after college (and/or some graduate programs). The goal of teachers is not to train students to perfect academic paper writing. It is to use the exercise of writing the paper (1) to give students an opportunity to consolidate and apply their learning through articulation (the process) and (2) to give the instructor some means of evaluating whether the student has done this, which in turn can help motivate the student to work harder at it. The paper is a tool, one of several possible means to an end. That end goal is the articulation of ideas that in turn produces transformative deeper learning. The ultimate goal of the humanities is to enable students to engage responsibly in their own formation in order to become happy and virtuous individuals and useful members of communities and the civic body.
Unfortunately, too often we emphasize the product over the process. The system is designed to prioritize the good paper that earns an A over the harder and riskier task of developing the kind of mastery and deep learning that leads to self-formation. Our educations system and institutions reward efficiency, and papers or written exams are, for faculty, relatively cost-effective. Education becomes transactional. Students submit papers in return for grades. Graduate schools and employers keep the pressure on, and it is often difficult for faculty members to separate their roles as educators from that of sorters and gate-keepers. The incentives are all stacked for placing product over process. They discourage the kind of intellectual risk-taking that often leads to real growth.
AI is seductive. It requires little effort and cost in return for an acceptable product. In my experience, AI, when given the proper prompting, can write good student papers. I have little doubt that it will soon be able to write great ones. In the meantime, though, a good paper might be good enough when it can save five, fifteen, or twenty-five hard hours of reading, thinking, writing, and revising. AI-produced text is like revelation, usually correct and easy to read and digest.
When it comes to writing (both for students and my own), I find AI to be useful in two ways. First, as a generator of utilitarian text, in which the writer’s voice is not valued. AI may eventually be better than most people at drafting contracts, legal briefs, organizational memos, and instruction manuals for putting furniture together (may that day come speedily). Second, I have found it useful as a conversation partner. I bounce ideas off it. I fed an earlier version of this essay into my bot and asked for critique. Some of the suggested edits were excellent, even as I fought to keep my own distinctive voice.
What I want in my classes, though, is to give students as many opportunities as possible to go through a process that will bring them deeper understanding. I try to emphasize that the paper or discussion post or oral comment in class are part of the same process. Perhaps perversely, I am especially drawn to moments of failure. There are few more powerful educatable moments than when a student watches their own false understanding crumble. When students use AI to create a better product rather than enhance the process, they shortchange themselves of such opportunities.
I want to convince my students that avoiding the seductive features of AI is good for them and that I value process over product. I think I have had partial success at doing this. For students focused on their grades and highly trained to focus on the product, though, this is a hard sell. Why should they trust me? They have reason to be skeptical. In my own mind, I am not fully clear how, when reading papers, to focus on evaluating the process over the product. I recognize that in terms of student learning a “bad” paper may be much more effective than a “good” one, but how does one reflect that in the grading? I wish I had a better answer than I currently do.
I still think back to my use of Cliff Notes for A Tale of Two Cities. I feel a tinge of guilt, but more so disappointment in myself. Some of that disappointment springs from avoiding the opportunity of engaging with a great work of literature (it is still on my reading list). More of it, though, comes from my inability at that age to discuss my situation with the teacher, with the hope that we could find a solution that better preserved my learning and integrity. That was a failure on my part, but one that I continue to learn from. When it comes to AI use, I hope that students do not make a similar mistake – and miss the harder, more valuable work of learning for themselves.
