One of my family’s prized possessions is an Amish-made table fashioned from solid cherry wood, a souvenir from the decade we lived in the Amish country of northeast Ohio. It’s a thing of real beauty: The hand that made it was sure of its own skill.
We loved that table so much we ordered a bedroom set from the same maker. When it was delivered, two strapping Amish teenagers gingerly lifted each massive dresser from the truck and carried them into our house. They were not buff but lean and fearsomely strong. I’ve had to move those dressers several times, and while a crew of ordinary men have to break a sweat to move them, the two Amish teens handled them as though they were nothing.
The Amish strangely said no to machine technology. They oddly said no to the internal combustion engine. But in doing so, whatever their reason, they held onto a tenacity of bodily being that we in vain try to recreate on our treadmills and our workout stations.
When we think of the Amish, we might get hung up on their hypocrisy, how they rely on non-Amish to drive the delivery truck or to build their webpages to advertise their furniture. But when we do that, we miss the point. They don’t just opt out of machine technology, more or less successfully. They also opt in to the excellence of handicraft and good old-fashioned bodily work, which forges and fashions their strong communities.
We are at an Amish moment regarding AI, for this new technology is not just one more incremental change.
Though I would love to have the craftsmanship or the strength of the Amish, I am probably willing to be a craftless modern if that means I can have the benefit of machine technology and the leisure it affords. After all, there are not, as far as I know, Amish universities. We machine-users might forgo handicraft, but in doing so we free up time to develop the truly enjoyable craft of thinking.
We are at an Amish moment regarding AI, for this new technology is not just one more incremental change. Like the advent of machine technology, it is a game changer, promising to affect us deep down. And as companies rush to integrate AI into all aspects of life, including the techniques of education, we might pause to consider the example of the Amish.
After all, we are human beings, not mindless robots. If we necessarily embrace whatever technological innovation happens to surface without asking how it might impact the goal of the task in question, we are failing in the work of discrimination proper to our natures. The critical question is this: If we wish to persist as genuine experts at the craft of thinking, to what extent—if at all—should we avail ourselves of AI assistance?
In many workplaces, AI has rapidly become an essential productivity tool. In the hands of the expert, such tools may be helpful. But my focus here is on the novice or student, the one who lacks knowledge. The goal of education is to lead a novice up out of ignorance and into expertise. What’s good for the expert may not be good for the novice and vice versa.
My argument in what follows is simple. The goal of education is to foster expertise in thinking. But AI hinders rather than helps this development. Therefore, AI has no place in education.
When “Help” Becomes Harm
When my children were young, I knew I could either tie their shoes right then and there, which was the quickest solution, or I could take a bit longer to teach them for the umpteenth time how to tie their own shoes; in the long run, though, that extra effort and patience was the only way they could learn to do something for themselves. If the parent always hovers and never challenges, never leaves their kids to struggle to do something for themselves, then their kids are going to turn out to be the ones always walking around in a panic every time their shoes become untied.
The philosopher Martin Heidegger speaks about the difference between leaping in and leaping ahead. When we leap in, we solve a short-term problem for people but in doing so create a long-term dependence, for we don’t teach them to solve the problem on their own. When we leap ahead, we show them how to handle the situation; we set them free for the future.
Help cannot be constant, or it fails to be help and becomes permanent dependence.
Education has always been about leading the student out of the bondage of ignorance to the liberation of illumination, to move from the privative state of the novice to the formed state of the expert. To the extent to which, in the name of assistance, we allow companies to create dependencies, we will have taken something successful and made it less effective, less efficient.
Consider the present effect of AI on the work of cultivating students. If professors share an article with their students who view it on a PDF reader, the program will helpfully, but also a bit pathetically, say, Boy this is a long document. Can I summarize it for you? And students—who can hardly read today as it is—might find it difficult to decline the offer.
AI companies constantly and cleverly wish to save the students the trouble of summarizing. But that trouble is not an impediment to their education; to be able to make sense of a complex topic by teasing out and eliciting hidden patterns of meaning is one of the fundamental skills of being well educated.
AI companies sell their product as the always-available tutor or professor, apparently not realizing that an always-available tutor or professor is no teacher at all. Help cannot be constant, or it fails to be help and becomes permanent dependence.
No professor, welcoming a student into office hours, tries to get the student to cede more and more to the agency of the professor; it is quite the opposite. The contest of office hours pits students against the wits of the professor; cunning but misguided students try to lure out of the professor a thesis statement or a thought they can then pass off as their own, and the professors, channeling their inner Socrates, use the opportunity to have students find, to their delight and surprise, that lurking down inside of them was a thought they had yet to discover, the discovery of which fills them with surprise, delight, and satisfaction.
Good office visits end with students thinking they didn’t gain much by stopping by. They were set free to think for themselves.
But AI is like an inexperienced adjunct on a one-year contract at a mismanaged university: someone desperate to be liked immediately by every student. “You want me to summarize the reading? Sure thing. Can I also brainstorm some topics to save you the trouble? Okay. Can I also outline the paper too? Yes? Say, why don’t I just write it for you, because I know how much you are struggling with it?”
What AI promises in education is education’s evaporation. To the extent that we embrace AI, we change the product of education from human liberation through thinking into human dependency through technology. What is difficult today will, on this path, become impossible tomorrow. Help! I no longer find the craft of thinking difficult but impossible; now I cannot do it for myself at all.
AI is making us learn a new vocabulary: “un-skilling,” the eroding effects of AI on expertise, and “never-skilling,” the eroding effects of AI on training. The student on AI never skills up; as such, the student on AI is no student at all.
The Aim of Education: To No Longer Be Necessary
If AI is a tool, what is the problem in education that AI can fix? Thinking is hard. There is no doubt about that. But the difficulty in education is something that no one can take away from the student without taking education away from the student. It requires effort but the effort is essential to the aim; eliminate the effort and you eliminate the education.
Generative AI, however, wants to brainstorm, outline, and write for students, and thereby save them from the trouble of thinking. The illusion is that somehow because such work is done at the bidding of students, it somehow counts for them. But if it does not bring the student any closer to achieving expertise, it has no educational value.
Consider a parallel. Suppose a tech company drops by the art department to showcase a new generative pottery wheel, able to help with all stages of production. All you have to do is press the on button, select a design, and it does the rest for you. The artists would do well to opt out of the machine, however efficient it was in production, for it eliminated the one thing necessary: the joy of cultivating and exercising human excellence. It would turn development into decline.
Or suppose the university, wishing to help students get fit, replaces all the manual gym equipment with machines that lift the weights at the command of students. Or the music department, wishing to foster musicality, replaces all the pianos on campus with self-playing ones that play at the command of the music students. It’s not the same, because the point was really never a matter of effecting outward but instead inward change.
The AI assistant, by contrast, appears ready for a lifetime of dependency.
There may be forces in technology that cannot be fathomed or controlled, but we are not powerless, for we can always say, “No, thanks. I’d rather walk or take a buggy or make gorgeous furniture and mugs by hand.” Or in this case: “Thanks, but no thanks, AI device: I’d rather draw or wrack my brains. After all, I’m okay with effort, for it is only effort that yields excellence and only excellence that yields satisfaction and only satisfaction yields a sense of lasting self-worth.”
The goal of all education is to become obsolete, to give the students the skillset so that they no longer need their teachers. Graduation is just that: recognition that students are no longer students because they have proved they can think on their own.
The AI assistant, by contrast, appears ready for a lifetime of dependency. You are to be a subscriber of the AI device for life, and that means you are no student but instead a customer. There will never be alumni of the AI “university,” never a graduation, because AI is opposed to the essential work of any university: developing the craft of thinking for oneself.
As Heidegger noted, “In such help, the recipient can become one who is dominated and dependent, even if this domination is a tacit one and remains hidden to him.”
Joyfully Built of Substance
Amish furniture does not just look good: It is good. There is no slick veneer laid atop a rickety substructure. Rather, the appearance is brought forth from the best natural materials. The result, as my Amish shop owner boasted, will last for two generations or more.
The student who wants to embody the wisdom of the ages, to acquire the skill of a master craftsperson, would do well to learn from the Amish: If we seek excellence in writing and thinking, some technological shortcuts must be scrupulously avoided.
Without the crutch of AI in school, graduates will stand out for their confidence and sense of the whole. They will not be appendages to their devices but human beings who can together face the puzzles and perplexities of the world. They will be able to evaluate the technologies in use because they are positioned to know, from the inside, just what it is that they are supposed to imitate. As knowers and producers, they will have an expertise with the device that a mere user can never have.
It is not really a matter of saying no to AI but a matter of saying yes to the real development of the craft of thinking.
In short, students who have “gone Amish on AI” will be better positioned long-term, and after they graduate and enter the workforce, they will be able to say why they are better positioned and be able to do so even without anyone—or anything—telling them what to say. They will be experts surrounded by perpetual novices.