Education Must Become Analog

Education in the AI age is being rebuilt in the wrong place. A student sends her essay to ChatGPT and gets it back smoothed out. She reads what’s there now, and it sounds smarter than what she’d written. She deletes her sentence and adopts the foreign one. What she can’t see: whether the foreign sentence is closer to the material or further from it; whether the smoothing sharpens it or flattens it; whether anyone is contradicting her at all, or whether she simply watched herself adopt it. She doesn’t know what she doesn’t know. And exactly inside this not-knowing sits the question that gets lost in every curriculum today: in what sequence a human being has to learn, for a tool to become a tool for them at all.

#The Sequence

The position can be stated briefly. Education has to be reset to analog, to a large degree. Not a little, not here and there, not “in addition to the digital,” but as the basic practice. Architecture is the clearest case. No computers at the start — the hand on paper instead. Old textbooks in old styles. The pencil that draws a line whose direction the person drawing chose themselves, not one a program suggests. Only once this practice sits in the body, in the eye, in the hand, in the imagination, is the computer allowed to join in. Then, and only then, does it become what it’s meant to be: an accelerator for a mind that’s already been trained.

What’s exposed in the architecture example holds everywhere. Whoever has mastered a language can use a translation program; whoever hasn’t is mastered by the program. Whoever has learned to calculate uses a calculator; whoever can’t calculate takes the display for the answer. Whoever has formulated texts themselves can be smoothed by a language model; whoever has never formulated anything has their text dictated to them and doesn’t notice. The asymmetry is always the same. It has nothing to do with the tool. It has to do with the one wielding it.

#Proof by Example

A small incident shows what’s meant. Whoever has spent a while at home in philosophy regularly catches language models at false attribution. An AI puts a sentence in Nietzsche’s mouth that he never wrote, presents it in a matching style, with a plausible-sounding citation. And it really does sound like Nietzsche. Whoever has read the late Zarathustra themselves hears the difference. They hear that this sentence rolls too smoothly, that the tension usually present there is missing, that the image is too unambiguous. Whoever hasn’t read it doesn’t hear this difference. They adopt the quote, copy it down, and have put a forgery into circulation without knowing it.

This small case is the entire matter in a single image. An analog philosophical education, carried out on printed books that no language model ever wandered through between the lines, leaves behind a benchmark inside the reader. It’s this benchmark that recognizes the forgery. An education that begins with the language model leaves no such benchmark behind, because the model itself was the source of what it should have been compared against. AI without an analog basis doesn’t work to reduce information — it works to increase hallucination. It multiplies plausibilities against which there is no outside standard.

#Humboldt, Schiller, Goethe — and the Picture We’ve Forgotten

Education in the classical sense didn’t mean data-feeding. Wilhelm von Humboldt (1767–1835) grasped it as the “highest and most proportionate cultivation” of the human faculties into a whole (cf. Humboldt, 1793). For him, a university wasn’t a knowledge-dispensing counter, but a “community of teachers and learners” (cf. Humboldt, 1810) — a relationship between people in which something is passed on that can’t be deposited in any textbook or in any model alone. Friedrich Schiller (1759–1805) expanded this picture in the Letters on the Aesthetic Education of Man (Schiller, 1795) with the play drive: that point between sensuousness and form where a human being first becomes free and whole. In Wilhelm Meister’s Apprenticeship (Goethe, 1796), Goethe called what education aims at self-activity. Bringing something forth yourself is the only form in which what’s learned is genuinely made one’s own. Confucius (551–479 BCE) saw lifelong learning as a practice that begins with the clarification of concepts and finds its meaning in the matured judgement of the noble person. All four lines say the same thing in four languages. Education is the shaping of a faculty on material that can be experienced bodily. It is not the accumulation of output.

What we call education today has little left of this line. Reading has become scrolling. Writing has become typing with autocorrect suggestions. Drawing has become selecting from a template. Listening has become playback at increased speed. The trace that a practice leaves in the body drops out, and with it the benchmark against which a digital result could be checked as right or wrong.

#Maieutics Needs an Empirical Basis

There’s an old form in which the creation of this benchmark is named. Socrates called it maieutics, the art of the midwife. He didn’t mean a question-and-answer technique by this, but touching your own empirical basis: what you actually think, not what you say you think. Maieutics delivers into the world what’s already laid down inside the learner. If nothing is laid down, it delivers nothing into the world. It needs material. It needs the substance that was read because it was read, not because a summary reminds you of it. It needs the line that was drawn because it was drawn, not because a program output it.

AI must not replace the maieutic act, or a foreign, smoothly formulated movement takes the place of your own movement of thought. What you then bring forth isn’t what had grown inside you, but what fit the question inside the model. Linguistically, the substitution can’t be told apart — but existentially, it’s the difference between education and its imitation. Whoever has carried out this process themselves knows this. Whoever has never carried it out mistakes the echo for an answer.

#What This Isn’t

This demand isn’t Luddism. Machines don’t get smashed for being machines. It isn’t nostalgia either. Nobody wants the mimeograph back. This isn’t about a bygone schoolroom world, but about a sequence that’s the same in every era. First your own practice, then the tool. First the hand, then the program. First the sentence you’ve read, then the search query. An analog education reform in this sense is technology-mindful, not technology-hostile. It takes the tool seriously enough to give it the place where it stays a tool. Whoever reverses that place no longer has a tool — they have a substitute for the person who was supposed to wield it.

#AI as Amplifier, Not as Teacher

The conclusion is a hard one. AI is an amplifier for trained minds, not a teacher for untrained ones. The same software, placed in front of an architect who has designed spaces by hand, accelerates their work. Placed in front of a student who has never drawn a line, it replaces their training. The same AI that is a research intern to a well-read philosopher is, to a student without any reading behind them, a source of sentences that can’t be checked. The difference doesn’t lie in the tool. It lies in the sequence.

Whoever takes this seriously builds education differently. Not “less digital,” but in two clearly separated phases. Phase one: physical learning, with the means that have worked for two thousand years because they leave a trace in the body. Book, pen, chalkboard, voice, someone across from you. Phase two, much later, after years: the tool that lends speed, that spares you routine, that smooths language, that sorts material. Phase two without phase one is not education. It’s the complete handover to a language surface that no longer has any outside to it. It’s exactly out of this situation that the state described as a diagnosis in the lexicon entry on AI psychosis arises. Whoever has no analog basis can’t contradict the language surface, because they lack the very thing that would contradict it.

That’s the position, stated calmly, without an uproar. It isn’t a polemic against technology. It’s a statement about what a human being has to be, for a tool to stay a tool. And it’s a statement about what education achieves, when it’s still what it was: the slow emergence of a faculty that arose inside the learner themselves, and by which they later recognize whether what a machine puts in front of them is true.

#Sources

Goethe, J. W. von (1796). Wilhelm Meisters Lehrjahre (Wilhelm Meister’s Apprenticeship). Berlin.

Humboldt, W. von (1793). “Theorie der Bildung des Menschen” (Theory of the Cultivation of Man). Fragment.

Humboldt, W. von (1810). Über die innere und äußere Organisation der höheren wissenschaftlichen Anstalten in Berlin (On the Internal and External Organization of the Higher Scientific Institutions in Berlin). Berlin.

Kirchhoff, G. (2026). “Beitrag beim Symposium zur Frage nach Künstlicher Intelligenz und Bewusstsein” (Contribution to the Symposium on the Question of Artificial Intelligence and Consciousness), Berlin, April 25, 2026, Position P11 [69:00–69:45] and Position P3 [34:30–35:15].

Konfuzius (ca. 500 BCE). Lunyu (Analects).

Schiller, F. (1795). Über die ästhetische Erziehung des Menschen in einer Reihe von Briefen (Letters on the Aesthetic Education of Man). Tübingen.