Does artificial intelligence render a Ridgeview education antique or indispensable?
A blizzard of essays and articles have appeared of late pertaining to the impact artificial intelligence is likely to have on education. Most of these articles are responses to the current state of Large Language Models (LLMs), but the difficulty in writing about these developments, aside from the problems with prognostication more generally, is that the transformer architecture, statistical prediction, datasets and AI training, and algorithms are changing so rapidly that what was true of AI’s implications three months ago are no longer entirely relevant. What does not change as rapidly are the philosophical questions that have endured for millennia.
On the one hand, the seductiveness of AI makes almost any human endeavor dubious. It is like the Irving Berlin song Anything You Can Do, except that here it is that AI can do it better, or so it seems. What first appeared as a liberation becomes an anxiety about what remains worthy of human attention and effort, and existential questions about purpose and meaning creep in. It seems as though we are inadequate and obsolete in a world of our own devising. The knowledge we have spent decades acquiring and fine tuning is enhanced in seconds with a simplistic prompt. What can we contribute that is original or worthwhile? More to our point, if our lives already seem to be of questionable value to an uncertain end, what are the prospects for our children? Specifically, what are the prospects of their being human? If being a teacher or a student is called into question, does how we understand what it is to be human separate from what has historically been meant?
Other questions abound. What is it to be a human being? How does the undeniable reality of AI alter the content, craft, and culture of teaching? Do we have a surplus of humanity? Who is master and who is man? The old philosophical questions persist. So, too, do the eternal or accursed ones. Is classical education well positioned to adapt?
Recently, I had a conversation with someone who was relating a story about working in public education. He related that he had initially been optimistic about AI acting to supplement learning, and providing meaningful and cogent tutoring for students struggling with a subject. He thought of it as a more intelligent and individuated version of Khan Academy complementing what was happening in his classroom. Three years later, he was entirely pessimistic. He said that teachers were using AI to invent lesson plans, discussion questions, vocabulary quizzes, and essay topics, and that students were using AI to produce their work, and that teachers were in turn using AI to grade this work. The only thing that had been evaluated in this was the AI. The performance of education continued, but it had lost all of its content, substance, and purpose. It is similar to what is probably the most famous line from C.S. Lewis’ Abolition of Man: “In a sort of ghastly simplicity we remove the organ and demand the function. We make men without chests and expect of them virtue and enterprise. We laugh at honour and are shocked to find traitors in our midst. We castrate and bid the geldings be fruitful.” What we wanted from learning is what we have removed from it by allowing students unfettered access to it.
At Ridgeview we have long said, and we are quoting from Aeschylus, that “we learn by suffering.” There is real profundity in that statement, but it is learning as understanding—not learning as the recitation of knowledge. The first problem with student’s access to AI is that they avoid the struggle that forms their minds. The grammar-stage work may seem tedious as we parse, copy, compose, and eventually proceed to a logic stage wherein we form arguments, identify fallacies, and then eventually invent, arrange, and speak out in our own voice. The seductiveness of AI robs us of developing a well-trained mind; in fact, it robs us of our ability to use AI intelligently.
The second problem is the one most discussed by teachers, which is a catalogue of their worries about academic honesty: it becomes impossible to clearly assess what a student knows if all of their work is produced offsite. The struggle becomes optional, which in turn becomes its own character issue.
The third issue that prevails is that memory and attention suffer, and the sort of impromptu speech that occurs face-to-face weakens. People broadly lamented the social deficits that formed during the pandemic, but even they are minor compared to a kind of pedagogy that does not lean in on recitation, narration, common-place books, and participation in seminars. Thinner attention, shallow memory, diminishing patience, and less facility speaking publicly all compromise a school that claims to educate holistically and that values wonder, awe, intellect, imagination, prudence, and fortitude. What begins as intellectual diminution ends as a catastrophe of character.
The fourth problem is that the machine cannot substitute for this. It is not wise, moral, or even inherently truthful. The LLMs flatten disputations and reflect the biases of their training. Truth, Beauty, and Goodness are enormous questions, but they have the benefit of being genuine questions; that is, questions without final, definitive, and uncontestable answers. Students should not be brought up believing that judgments can be outsourced. Ridgeview said this aggressively when judgment was outsourced to other humans; but, it must do so all the more urgently when it is being outsourced to machines.
The fifth problem is that because the AI remembers that you have a dog and are allergic to walnuts and much else, it has the sort of familiarity we might associate with a person. It is not. You cannot be known and loved by an AI. Students ought to belong to a human community. When we assert that school is part of socialization, we do not mean that it is the school’s right to indoctrinate children into an ideology or wedge them against their parents. We mean that students gain practice engaging in civil discussions about serious and oftentimes controversial matters. They have to learn how to be their best selves in a world that is not always good, accommodating, or wholesome. Their teachers help to foster this through Socratic questioning, providing good models of behavior, correcting vice, and praising virtue. The teacher and the student are one type of association, and in this association, both teachers and students become more human.
AI does, however, change some things about how a classical education is communicated. First, AI might legitimately strengthen a lesson or one’s assessment of an argument’s strengths or weaknesses. This is as true of the teacher’s lecture as it is of the student’s paper. With this said, both student and teacher must begin with what is in their own minds. There is academic honesty, but there is also the professional honesty, which involves being accountable for one’s work being one’s own work. Ridgeview does not offer a scripted curriculum. There are no shortcuts here for either students or teachers.
Intimacy of instruction becomes more important in an age of AI rather than less. Students are to be praised for doing well, but all of them are to be pushed. That can only happen from a teacher who has a regard for their development reacting empathetically, organically, and in real time. To quote from the St. Augustine poster on my classroom wall, “The turbulent have to be corrected, the faint-hearted cheered up, the weak supported…those engaged in quarrels reconciled…the good to be encouraged, the bad to be tolerated, all must be loved,” and AI cannot do this. To be known and loved is the sort of work people must do.
Teaching students to write well will forever be a mainstay of what Ridgeview does, but recognizing its potential limitations in an age of AI means a greater emphasis on bluebook exams, viva voce, classroom participation, looking at the marginalia in student’s books, and prioritizing conversations with students. Additionally, asking big, substantive, discursive questions, and prompting them to think without giving them answers to recite. This is what we already do when we ask them thesis questions like “What is the nature of the heroic?” or “What is the place of religion in the life of man?”
Finally, AI only answers what one knew to ask. This is Meno’s Paradox: you cannot search for what you do not know at all, because you would not recognize it. A mind must be trained to ask insightful questions, and no generation entirely inherits the progress made by its predecessors. We may stand atop the shoulders of giants, but some things must be learned anew by each successive generation. And, there is real value in practical wisdom. The philosopher Michael Oakeshott addresses this obliquely in his essay Rationalism in Politics by relating a Chinese illustration demonstrating that knowledge cannot be reduced to technique. Practical knowledge, of anything, which includes character, tying knots, writing, doing math, painting, or playing the piano, exists only in use, cannot be formulated, and is acquired by apprenticeship or deep study. Being told is insufficient; you have to do a thing to really know a thing. Life gives us opportunities AI does not.
This is not my final testament or all of my comprehensive thoughts about AI in education. I certainly could have razzle-dazzled readers with more academic and highfalutin fare, but it is written to assert that classical education is well aware of the challenges posed and is not resorting to Ludditism. If it seems agnostically calm about the prospects of AI’s impact on education, it is because classical education is largely proposing to continue doing what it has mostly always done. And, while the answer might be construed as self-interested or biased, Ridgeview seems more necessary and more indispensable than ever. Becoming human is still a sought-after pursuit even in an age of machine learning. Even with omniscience at our fingertips, we must still be, and what and how we will be remains the purpose of education.
D. Anderson
Headmaster
