Institutionalized Education as Cognitive Offloading

Schools have been the epicenter of cognitive offloading since the 1920s. Learn what I tell you. Think this. Don't challenge authority. Cram for the test. Stay within the frame.

May 19, 2026

A student turns in an essay her teacher knows she did not write. The five paragraphs are correct. The thesis sits in the right place. The transitions are smooth. The conclusion restates the introduction without saying so. The rubric is satisfied. The grade is a B+.

Everyone in the room knows what happened. No one knows what to do about it, because the honest answer is the one nobody wants to say.

The AI crisis in education is not that students suddenly learned to outsource their thinking. It is that schools spent a century teaching them to do exactly that — and then panicked when the outsourcing got too efficient.

Plato saw the shape of the problem twenty-four hundred years ago, and he was wrong about it in exactly the way we are now wrong about ChatGPT.

Thamus was right and wrong

In the Phaedrus, Socrates tells the story of Theuth, the Egyptian god who invents writing and presents it to King Thamus as a gift to humanity. Thamus refuses the gift. Writing, he warns, will not strengthen memory — it will destroy it. People will stop knowing things and start looking them up. They will appear wise without being wise. They will become “the disciples of many things, and will have learned nothing.”

Thamus was right about the mechanism and wrong about the verdict. Writing did hollow out the kind of memory the oral tradition had cultivated. The Homeric bards who could recite tens of thousands of lines are gone, and they are not coming back. But we got something in return — a civilization that could accumulate knowledge across generations rather than losing it with each death. We made the trade. It was the right trade. We just had to give up a particular kind of human capacity to make it.

This is the question Thamus actually pressed, even if he did not know he was pressing it: which cognitive functions can we afford to offload, and which ones, if offloaded, hollow out the human capacity we cannot replace?

Every generation answers this question in practice, by what it builds. Institutionalized education is one of the most ambitious answers ever attempted.

The first great offload

We tell ourselves that schools exist to develop children. That is true, but it is not what built them.

Mass compulsory schooling — the model nearly every country on Earth now operates — is roughly 150 years old. The 1870 Forster Education Act in England, the Prussian reforms before it, the American common school movement, the wave of national education systems that swept the industrializing world between 1870 and 1920 — none of these emerged because someone discovered childhood. Francisco Ramirez and John Boli’s comparative work on the global diffusion of mass schooling makes this plain: states adopted compulsory education in waves not because pedagogy demanded it but because state-building did. Andy Green’s Education and State Formation tracks the same pattern across England, France, Prussia, and the United States.

What did the institution do, structurally?

It offloaded socialization from families, guilds, and churches onto a centralized state apparatus. It offloaded the transmission of basic literacy from parents who often could not read onto trained specialists. It offloaded time-discipline, deference to written rules, and tolerance for boredom from the household onto a building with a bell.

This was a cognitive and social offload at civilizational scale, and by most measures it was a triumph. Literacy rates that had hovered in the single digits for most of human history climbed past 90% in a century. The institution worked.

Then it kept going.

By the 1910s and 1920s, schooling had taken a second turn — one that matters more for this essay. The institution that received those families’ offload in 1870 was still, in important ways, a human-judgment-driven institution. The teacher decided what to teach today. The headmaster decided who advanced. The community decided what was worth teaching. Then, in roughly fifteen years, that judgment was moved out of human hands and into systems.

Ellwood Cubberley, dean of Stanford’s School of Education and the most influential American educational administrator of his generation, wrote in his 1916 Public School Administration: “Our schools are, in a sense, factories, in which the raw products (children) are to be shaped and fashioned into products… The specifications for manufacturing come from the demands of twentieth-century civilization, and it is the business of the school to build its pupils according to the specifications laid down.” The line is famous because it was offered proudly, not in protest.

Raymond Callahan’s Education and the Cult of Efficiency documented what happened next. Curriculum decisions migrated from teachers to central offices and the new class of professional administrators Cubberley was training. Evaluation migrated from teachers to standardized instruments — the Army Alpha of 1917 proved you could rank-order 1.7 million humans cognitively, on paper, at scale, without ever talking to them, and within a decade the same logic was inside every school district. Sorting migrated from local discernment to tracks set by test scores. Time itself migrated from human rhythm to the bell schedule and the Carnegie Unit. The purpose of school migrated from the local community to a national-industrial frame.

This was the deeper move. The first wave of mass schooling offloaded from families to the institution. The 1920s offloaded from human professionals to systems. Teachers were the first deskilled workforce inside the school, before the students were. The whole apparatus learned to produce cognitive judgments — about what to teach, what was learned, who belonged where, what counted as done — without the cognitive labor of a human exercising professional discretion.

This is the architecture we are still running. It is the architecture AI is now perfecting.

What we taught the students to do

Schools have been the epicenter of cognitive offloading since the 1920s. Learn what I tell you. Think this. Don’t challenge authority. Cram for the test. Stay within the frame. That is not a caricature; it is the operating system the Cubberley generation installed, and we have been running variants of it for a century. Each upgrade — standards, accountability, NCLB, Common Core, evidence-based curriculum — reinforced the architecture rather than questioned it. The frame tightened. The voice telling students what to think got louder, more confident, more centralized, more “scientific.”

Inside that frame, the student’s job has been clear and consistent: receive, repeat, perform, and do not ask whether the frame itself is the problem.

Consider what the standard apparatus actually trains.

The five-paragraph essay teaches that thinking has a known shape, the shape is given by authority, and conforming to it produces the grade. The student is not asked whether the question deserves five paragraphs, or three, or fifty. The cognitive work of form selection — arguably the central act of any serious writer — has been outsourced to convention.

The rubric tells students in advance what counts as good. This is defended as transparency, and it is transparency, but it is also an instruction to stop asking what good is. The cognitive work of evaluation has been outsourced to a document handed out at the start of the unit.

The textbook presents knowledge as a settled inheritance to be received, not a contested terrain to be navigated. The cognitive work of epistemic vigilance — knowing which claims to trust, which sources are interested, which framings smuggle assumptions — has been outsourced to publishers.

The standardized test rewards the production of pre-correct answers under timed conditions. The cognitive work of defining the problem — the hardest and most valuable thing a human does in any field — has been outsourced to test designers.

And the grade itself is an offload. The teacher does the evaluation; the student receives the verdict. Whether the student agrees with the evaluation, whether she could defend her own work against it, whether she has any internal sense of what is good independent of the mark — none of this is asked. Her job is to receive and adjust.

Paulo Freire named this in 1968. In Pedagogy of the Oppressed he called it the banking model — students treated as containers into which the institution makes deposits, then later checks the balance. Freire’s critique was political; he was watching the banking model train the colonized into docility. But the structure he described is also a structure of offloading. The teacher does the thinking. The student receives the deposit. Half a century later, the system Freire diagnosed has only deepened — and now the depositor can be a chatbot.

We built an institution that ran on offloaded cognition for a century and a half, and then we acted surprised when students offloaded their essays to ChatGPT.

They were doing what we taught them to do. We were just used to being the recipient.

Schoolishness

I want to press this harder, because the offload of cognition is the visible part — the part you can name in a faculty meeting without losing the room. Underneath it is a deeper offload, and this one is harder to look at directly because the entire apparatus depends on us not looking at it.

The institution does not only offload how students think. It offloads what they are permitted to think about.

It tells them which questions are worth asking and which are not. It draws the perimeter of acceptable inquiry — what is on the test, what is in the curriculum, what is in the unit, what falls under “we’re not going to get into that here.” It tells them which books are on the shelf and, in a rising number of districts, which books have been pulled from the shelf because someone decided the students should not encounter them. It tells them what counts as evidence, what counts as a good source, what counts as a defensible position, what counts as too political, too religious, too sensual, too uncomfortable, too dark, too unsettled.

It tells them, by the architecture of the school day itself, what is worth their attention: forty-five minutes of math, forty-five of English, forty-five of science, a thinner slice of art, almost nothing of philosophy, civics reduced to a test prep unit, ethics absent entirely, the questions that organize a human life relegated to “electives” or to whatever the kid finds outside the building. The schedule is a curriculum. The curriculum is a statement about what is real and what is decoration.

It tells them what their future is for. The college-and-career pipeline is the master frame, and the master frame channels the student’s emerging sense of self toward participation in a specific economic order — productivity, earnings, credentials, employability, fit. Students who do not orient themselves around that frame are diagnosed as disengaged, at-risk, in need of intervention. The institution offloads from the student the cognitive and existential work of deciding what a life is for and replaces it with a pipeline diagram. By the time the student is sixteen the question who am I becoming and why has been quietly substituted with the question what are you applying to. These are not the same question. One is the work of a soul. The other is the work of a guidance counselor.

It tells them which forms of disagreement are tolerated. A student who pushes back on a teacher’s framing in the wrong way at the wrong moment is not having their argument engaged; they are being managed. A student who refuses an assignment on principled grounds is not having a moral seriousness recognized; they are being marked down. The institution preserves the form of disagreement — “controversial issues” units, opinion essays with prompts — but it controls the terms under which disagreement is permitted, and a student who disagrees outside those terms learns quickly that the system does not actually want their mind.

It tells them what to want. The reward structure of the school — grades, honors, advanced tracks, GPA, class rank, recognition, college admission — substitutes an external scaffolding of desire for the internal work of figuring out what one actually cares about. Twelve years inside that scaffolding is enough to permanently confuse the two. Many adults never recover. The high-achieving student who arrives at college unable to answer the question what do you want to study without first asking what looks best on the transcript is not a failure of the system. The system did its job.

This is what Philip Jackson, observing classrooms in the 1960s, called the hidden curriculum — the lessons in deference, punctuality, queueing, gendered participation, and acceptance of evaluation that no syllabus names but every classroom teaches.

It is also what the Black unschooling writer Akilah Richards, in Raising Free People, has been calling, more bluntly, schoolishness — “the things that happen throughout our schooled lives,” the residue of institutionalization that remains in a person’s relationship to their own mind long after the schooling has ended.

Richards’s term is sharper than Jackson’s because it names the thing as a quality of the person rather than a feature of the building. The offload is not only of cognition. It is of orientation. The student is taught not just to look outward for the answer, but to look outward for the question, for the criteria, for the perimeter of permissible thought, for the shape of a worthwhile life, for the verdict on whether their inner life is doing it correctly.

This is also what Ivan Illich saw with unusual clarity in Deschooling Society in 1971 — that the deep effect of schooling is not what it teaches but what it conditions: a population that has learned to confuse being credentialed with being competent, being taught with being learned, and the institutional voice with the voice of reality. And it is what Samuel Bowles and Herbert Gintis, in Schooling in Capitalist America (1976), called the correspondence principle: the structure of school relations mirrors the structure of workplace relations, not by accident but because that is the function of school in a capitalist order. The whole apparatus is calibrated, knowingly or not, to produce the kind of person the economic system can use.

And here is where the AI conversation gets clarifying in a way that almost no one wants to name.

An institution that has spent a century and a half offloading from students the cognitive work of deciding what is worth knowing, what is worth arguing, what books to read, what questions to ask, and what futures to want — is not in a strong position to object when those same students offload the residue of the work onto a language model. The model is the logical endpoint of the apparatus.

We taught students that the answer to what should I think lives outside them. ChatGPT is just a more responsive version of the place we told them to look. We told them their job was to produce the response the institution wanted; AI is exquisitely good at producing the response the institution wants. The students learned the lesson. The lesson was the offload.

This is the part that I think educators have the hardest time saying out loud, because saying it out loud requires acknowledging that the crisis is not a crisis of integrity or of effort or of technology. It is a crisis of what the institution was for. If the institution was for the cultivation of independent minds capable of self-government, the AI moment is a catastrophe. If the institution was for the production of compliant cognitive workers fitted to an industrial economy, the AI moment is just an upgrade in the production line. We have built a system that — for most students, in most classrooms, most of the time — works much better at the second purpose than at the first. We pretended it was doing the first. AI has called the bluff.

What it did to the teachers

There is a symmetry here that the standard reform conversation refuses to look at.

The same institutional logic that trained students to offload thinking outward has spent the last forty years working, with remarkable consistency, to offload thinking outward from teachers too.

The vehicle is standardized curriculum. The argument is reasonable on its face: teachers vary, outcomes are uneven, and a child in one room should not get a measurably worse education than the child next door because she drew a less effective adult. So we standardize. We adopt a scope and sequence. We script the lessons. We align the assessments. We monitor fidelity of implementation. We narrow the variance.

The unspoken second half of that argument is that we are doing this because we do not trust teachers to make the decisions themselves. If we trusted them, we would not need to script them. The scope and sequence is a vote of no confidence dressed as equity. The institution is saying to the professional standing in front of the children: we will do the thinking. You will deliver it.

There used to be a phrase for this that the reformers said proudly: teacher-proof curriculum. The phrase has fallen out of fashion. The product has not. It has only been renamed — “evidence-based,” “high-quality,” “aligned,” “coherent.” These are good words. They are also, in many implementations, the same offload: the cognitive work of deciding what a particular child in a particular room on a particular morning most needs to think about taken out of the teacher’s hands and handed to a curriculum designer in another state, often working for a publisher, often years before the child was born.

Michael Apple has been making this argument for forty years. He calls it the https://api.pageplace.de/preview/DT0400.9781136634789_A23859914/preview-9781136634789_A23859914.pdf of teaching — the same word industrial sociology uses for what Frederick Taylor and his stopwatch did to craftsmen in the 1910s. The deskilling is structural. It is not a side effect of bad implementation. It is what the design does.

The cost is not borne by teachers alone. It is borne by every student who has encountered a teacher reading from a script with the lights of judgment dimmed, who has watched a thoughtful adult shrug and say the pacing guide says we have to move on, who has asked a real question and received a curricular answer. Children can tell when the adult in front of them has been deskilled. They respond to it by deskilling themselves further.

This is also why AI lands in classrooms the way it does. A teacher whose professional judgment has been displaced for a decade by a centralized curriculum apparatus is not in a strong position to push back when a centralized AI apparatus arrives to displace what remains. The institutional muscle for no, I will decide what this child needs has atrophied. Teachers were already being told to offload their thinking to the program. Now the program is offloading itself to the model. The students are doing what the teachers were doing. The teachers are doing what the institution had always done.

There is no version of this where children learn to think under pressure in front of a teacher who has been instructed not to.

The credential collapse

Step back further.

The whole stacked system of K–12 grades, transcripts, GPAs, standardized tests, degrees, and credentials is itself a vast cognitive offload — this one from society onto institutions.

When an employer hires a bachelor’s degree, she is outsourcing the judgment of competence to the institution that conferred the degree. When a graduate program admits a 3.9 GPA, it is outsourcing the judgment of intellectual quality to the previous institution’s grading. When a state licenses a teacher, a nurse, a lawyer, a doctor, it is outsourcing the judgment of trustworthy practice to a credentialing apparatus. The credential is a shortcut. It exists because individually evaluating every person for every position would be impossibly expensive. So we built a machine that evaluates once, stamps once, and lets the stamp travel.

This is the conflation Illich named half a century ago — schooling gradually replacing learning as the thing society could verify, until eventually the verification was the learning, at least for the purposes of the labor market. The credential became the substance. The stamp became the thing.

What AI is doing to that stamp is roughly what the printing press did to the scribe’s monopoly. The credential rested on a signal — that the holder could do something the unstamped could not — and AI is collapsing the signal across a widening range of cognitive tasks. If a paralegal’s core work can be done by an AI system at one-twentieth the cost and three times the speed, the credential does not vanish overnight. But its premium collapses. The wage was paying for a scarcity that no longer exists.

The institutional response so far has been institutional self-defense — moratoriums, bans, AI-proof assessments, return-to-blue-book proctoring, accreditation regimes that treat AI as a threat to academic integrity rather than as a hammer falling on the credential’s foundations. None of it works. None of it can. You cannot defend a credential whose signal has collapsed by tightening the rules of a test that no longer measures the thing.

The serious question is not how to preserve the credential. It is what we can certify after the credential. What becomes worth measuring when the old measurements no longer mark a difference.

Where the student cannot disappear

The question is not whether students will use AI. They will. The question is where the institution creates moments in which the student cannot disappear behind the tool.

This is the question the next decade of education policy will be organized around, whether the field knows it yet or not:

Which human cognitive capacities, if offloaded onto AI, end the thing that makes a citizen a citizen?

When you press on this, the answer is not abstract. It picks out a recognizable cluster of capacities — the same cluster, as it happens, that has been cultivated for two and a half millennia by one of the oldest pedagogical traditions in the West.

Real-time adversarial reasoning. Another person is sitting across from you, attacking your argument, looking for the weak link, pressing you on the assumption you tried to hide. There is no offload. The model cannot stand in for you in the cross-examination. You are responsible for the next sentence, and the next, and the one after that, and you have to produce them in a body that is being watched.

Ethical judgment under uncertainty. The data is incomplete, the stakes are real, the principles conflict, and you have to act anyway. No system offloads this for you, because the act of judging is what makes you the agent rather than the executor.

Accountability. Standing behind a claim. Owning it when it fails. Revising in public. This is not a skill the way “five-paragraph essay” is a skill. It is a posture toward truth, practiced in front of others, or it does not develop.

Defending ideas under cross-examination. This is the formative test the system has not figured out how to fake, because the test is the live performance of a mind, observable by other minds, in conditions where the production of the next utterance is structurally non-offloadable.

These are, not coincidentally, the capacities that form the spine of debate as a pedagogical practice — Dialogue, Evidence, Balance/Argumentation, Thinking Under Pressure, Ethics. The D.E.B.A.T.E. framework is not a repackaging of debate as something fashionable. It is an attempt to name precisely the cluster of capacities that cannot be put on the conveyor belt without ending the citizen.

It is also the most AI-resilient formative assessment we have. A live cross-examination is not gameable by ChatGPT. A debate round is not generated by Claude. An adversarial exchange in which the student has to think on her feet in front of a judge who will press her is a form of measurement that has survived every previous wave of cognitive offloading — the printing press, the calculator, the internet, the search engine — and will survive this one.

The reason is structural. In a debate round, there is no rubric standing between the student and the act. No five-paragraph form to hide behind. No answer key. No textbook. There is another human, watching, listening, pressing. The student is responsible for the next utterance. She cannot offload it. If she has not done the thinking, the room knows.

That is what we cannot afford to offload. That is what is left when the rest is gone.

Who steps up

The serious work begins after the diagnosis.

If institutionalized schooling was the great cognitive offload of the industrial age, and if AI has now exposed what that offload cost us, then the question is not how to preserve the system. It is what comes next — and who builds it.

Start with what is obvious and what the critique cannot pretend away. Students still need content. They need to know things. They need to have read books, encountered history, learned mathematics, understood the methods of science, struggled with ideas they did not choose, met traditions older than themselves. The argument has never been against content. The argument is against the delivery mechanism the industrial school built around content — the standardized curriculum, the pacing guide, the scripted lesson, the rubric, the receipt-and-repeat assessment cycle that trained students to relate to ideas as inventory rather than as terrain.

But there is a deeper version of this argument, and it is the one most reformers will not say out loud. It is not just the delivery mechanism. It is the obsession the mechanism rests on. It is the conviction that there is some right body of content most students need, that they should spend 8 hours a day on it, that we can specify it in advance, that getting the specification right is a solvable problem, and that the next reform cycle — better standards, better materials, better alignment, better implementation — will finally deliver it. This conviction is the engine of standardization. It is what keeps the apparatus running through wave after wave of evidence that it is not working. And it is held with a quasi-religious intensity by people who would not call themselves religious about anything else.

The obsession has three layers. The first is the belief that most students need the same content, even as they age — that what an seventeen-year-old in Boise needs to learn is also what an eleven-year-old in Atlanta needs to learn, which is also what an eleven-year-old in Honolulu needs to learn, and that the differences between them are noise to be filtered out rather than information to be honored.

The second is the belief that there is some magic content — that if we could just identify the right canonical knowledge base, the right cultural literacy list, the right sequence of skills, the right grade-level expectations, the students would learn it and the system would work. Every generation of reformers believes they are close to this list. Each generation produces its own version. None of the versions has produced the outcomes claimed for it. And the third is the belief that the adults who design the content know more than they actually do — about which knowledge will matter, in which order, for which children, for which futures, in a world none of us can yet see clearly.

None of these beliefs survives serious contact with what AI is now doing to the labor market, the information environment, or the production of cultural knowledge. The futures the standardized curriculum was designed to prepare students for are dissolving in real time. The content the curriculum was designed to deliver is now available to any student at any moment, for free, in any language, on demand. The adults specifying the curriculum cannot reliably tell you which capacities a person born in 2018 will most need in 2040, because nobody can. And yet the standardization conversation continues — at the federal level, the state level, the district level, the school level — as if specifying the content harder will finally make the system work.

It will not. The obsession with getting the content right is the institutional version of the offload the essay began describing. It is the apparatus trying to do the thinking that only the student, in conversation with adults who actually know her, can do. The standardized curriculum is not the answer to a difficult question. It is the institution refusing to ask the question.

Content does not have to be standardized (or only standardized) to be serious. In fact, the more standardized it becomes, the easier it is for AI to replace the human who delivers it. This is the structural irony of the moment we are in. A society that produces standardized minds — minds shaped by the same scope and sequence, evaluated by the same rubric, certified by the same credential — is producing exactly the population AI is best at substituting for. The factory model is not a defense against AI. The factory model is what AI is replacing. The more our students look like the average graduate of the average program, the more replaceable they are.

The way out is not less content but less standardization of content. Different schools, different traditions, different paths through the material, different combinations of subjects, different ways of measuring what a student has actually come to know and do. Variance is a feature, not a bug, in an age when the median performance of any cognitive task is about to be free.

And this is where the argument has to turn from the institution to the student.

There is a story the industrial school tells about students, and it is one of the most expensive lies in education. The story is that students are not naturally interested in learning, that interest has to be manufactured, and that the school’s job is to motivate the unmotivated through grades, gold stars, threats, rewards, behavioral incentives, and the long-term promise of credentials. The story is necessary for the system to function. If students were genuinely interested in what they were being asked to do, the entire apparatus of compliance — the bell, the hall pass, the late penalty, the missing assignment notification, the GPA — would be redundant.

The story is also false.

Students have interests. They have many of them, often passionately, often in defiance of every effort to redirect them. They have interests in music, in games, in social dynamics, in fashion, in sports, in politics they’re told they’re too young to care about, in identities they’re told they’re too young to claim, in injustices they see clearly and adults pretend not to. They make things, learn languages, master complex technical systems, debate online with strangers, build communities, teach themselves skills the curriculum does not offer. They do all of this while also being told they are disengaged from school. The disengagement is real. It is not a disengagement from learning. It is a disengagement from the industrial form of learning the school is offering them.

What the industrial school stripped out, more than any specific content, was agency — the experience of one’s own choices having consequences. Agency is not the same as motivation. Motivation is something teachers try to install in students. Agency is something the student already has, that the institution either makes room for or grinds down. The standardized classroom is a low-agency environment by design. The student does not choose the topic, the text, the pace, the format, the question, the audience, the timeline, the partners, or the standard of judgment. She is told all of it. After twelve years of being told, she develops what the industrial school then diagnoses as disengagement — and what is actually the rational adaptation of a person whose agency has been systematically denied. She has learned that her preferences do not move the system. So she has stopped offering them.

A school that develops agency is not the industrial school with better engagement strategies bolted on top. It is a structurally different relationship between the student and the work. The student helps decide what is studied, when, with whom, to what end, and by what standard. The teacher is no longer the dispenser of approved content but the more experienced practitioner working alongside her, raising the level of what she is trying to do, defending the discipline’s traditions where they matter, getting out of the way where they do not. The standard is not the rubric. The standard is whether the work is any good — judged by adults who have actually done the kind of work being judged, by other students whose own work has earned them the right to a vote, and, eventually, by the student herself, who has developed taste through enough exposure to know.

This is harder than the industrial model, not easier. It requires adults willing to give up control they could legally exercise. It requires institutions willing to admit that their authority over what a student studies is no longer self-evident. It requires the recognition that interest, once permitted to be real, will sometimes lead the student to places the curriculum did not anticipate — and that this is the point, not a problem. Most institutional resistance to student agency, dressed up in the language of standards and rigor, is actually resistance to the loss of adult control. It will have to be named and faced.

The students are not the obstacle. The students are the engine. The obstacle is a century of institutional muscle memory that does not know how to operate when the student is allowed to want something.

This points at what the redesign actually requires. Not all new tools. Not better technology in current classrooms. The redesign requires new practices — different ways of learning, organized around capacities the machine cannot replace, with the student’s own interest as the live current running through them.

None of this means there is nothing all students should learn. It is almost certainly true that some shared content matters, particularly in the lower grades — that a society which fails to teach every child to read, to do basic arithmetic, to recognize the rudiments of how the world works and how its own past was shaped, has failed them in a way that cannot be repaired later. The early years carry an obligation the rest of schooling does not.

But even there — even where a shared floor is necessary — the institution has to make room. Room for the student to pursue what she finds, room for the questions the curriculum did not anticipate, room for the work she chooses to do well beyond what was assigned, room for the interests the school has historically diagnosed as distractions.

Agency is not the reward for finishing the standardized portion. It is the soil the standardized portion is supposed to be planted in. A school that teaches a child to read but does not let her read what she wants has taught her the technique and revoked the point. A school that teaches the methods of science but does not let her ask questions the curriculum cannot grade has trained a technician and refused her the inheritance. The shared floor is real. So is the human standing on it.

This points at what the redesign actually requires. Not new tools. Not better technology in classrooms. The redesign requires new practices — different ways of learning, organized around capacities the machine cannot replace, with the student’s own interest as the live current running through them.

Some of them are old. Apprenticeship is old. So is the seminar, where students sit around a table and argue about a text. So is the workshop, where someone who knows how to do something teaches someone who does not, in person, over time, with judgment passing both ways. So is debate. So is the studio, the lab, the rehearsal room, the field trip with a guide who knows the territory. These forms cultivated humans for centuries before the factory model arrived. They worked then. They will work now. They were displaced by the industrial school not because they were ineffective but because they did not scale to mass literacy in the way the bell schedule did. The conditions that made the trade-off worth it are ending.

Some of them will be new, and we do not yet know what they look like. The next form of education is going to be invented in places the current system does not look. It will be invented by teachers who refuse the script, by parents pulling kids out of systems that no longer serve them, by debate coaches and writing instructors and music teachers and shop teachers who already know what non-offloadable work feels like, by employers who get tired of waiting for credentials that mean nothing, by community institutions remembering what they used to do, by students inventing things their teachers cannot grade. It will not look like a single national program. It will look like a hundred experiments, most of which fail, some of which root, and from which — if we are honest about what we are seeing — a different kind of institution slowly comes into focus.

The question is who steps up. Not who is appointed. Not who is funded by the next federal initiative. Who, in the actual places where students and adults meet, takes responsibility for the work that the industrial school can no longer pretend to do.

And the redesign cannot be optimized for the AI economy. That is the trap. If we treat school as a workforce-preparation problem, we will lose the argument, because AI is already a better workforce than most of our graduates. The redesign has to be optimized for the kind of human a person becomes inside it: someone capable of judgment, capable of taste, capable of ethical reasoning, capable of standing behind a claim, capable of caring about something other than her own credentialed advantage, capable of participating in a democracy that will not survive a population that has outsourced its thinking.

That is human flourishing in the AI age. Not preparation for the economy. Preparation for the life the economy is supposed to serve.

This is the work. It is not glamorous, it is not centralized, and it will not be funded by the people who funded the last hundred years of schooling. It will be done by people who notice that the children in front of them deserve better than what the system is currently producing, and who decide to build something else, often without permission.

The institutions that figure this out first — schools, networks, leagues, programs, communities — will not be the ones that defended the credential the longest. They will be the ones that asked, earliest and most honestly, what a young human is for.

Coda

Thamus was right about the mechanism. Every offload hollows out the capacity it replaces. The bard’s memory is gone. The mental arithmetic that filled Victorian schoolbooks is going. Cursive is going. Whatever cognitive practice the textbook trained — the patient receipt of an authoritative narrative — is going too.

The question is never whether to offload. We have been doing it since Theuth presented his gift, and we will not stop.

The question is what we hold back from the trade because losing it would cost us something we cannot recover.

Institutionalized education was the great offload of the industrial age. It worked. It also taught two generations to offload their own thinking onto the structures designed to evaluate them, deskilled the teachers who were supposed to know better, and built a credentialing apparatus that offloaded society’s judgment onto a stamp that AI is now removing the ink from.

We do not need to defend the old structure. We need to be honest about what it was, what it can no longer pretend to be, and what we will not let go of next.

Cross-examination. Live argument. Defense under pressure. Judgment in front of other minds.

A student, in a room with other students, defending an idea she has actually thought.

That is the part we keep.

Education Disrupted: Teaching and Learning in An AI World

Discussion about this post

Ready for more?