The Same Lever
A manifesto
For two thousand years, learning was a relationship. A person who knew sat with a person who wanted to know, and something passed between them that had no name — call it attention, call it craft, call it the slow transmission of judgment from one mind to another. Socrates had no curriculum. The medieval apprentice had no rubric. The child who followed a blacksmith or a midwife or a navigator learned by proximity, by imitation, by the gradual awakening of their own capacity to see what the master saw. Genius arose from obscurity because someone was hungry and someone else was paying attention, and the two of them made something that neither could have made alone.
Then we industrialized it. The system converted the mentor into a content deliverer and the workshop into a classroom. It replaced the question with the lesson plan and the apprentice’s hunger with the student’s compliance. It took the most human exchange we know — one person transmitting judgment, craft, and attention to another — and rebuilt it as a logistics operation. Content delivered. Assessments passed. Credentials issued. It sorted children by age. It divided knowledge into subjects. It built buildings to hold them and bells to move them and tests to rank them. It called this progress. It was a factory. It has been a factory for a hundred and fifty years, and the products of that factory have been people trained to sit, to receive, to reproduce on command, and to mistake this for learning.
Every reform since has rearranged the factory floor. New desks. New schedules. New technologies. New standards. And at each turn, the same quiet axiom at work beneath the surface, never stated but always operative: The time has come when education must discard all reference to the inner life of the learner. Its sole task is the prediction and control of outcomes; and curiosity can form no part of its method. No one wrote that sentence. No one had to. The system wrote it through a hundred and fifty years of practice — through evidence-based learning that produces evidence of its own existence and calls it proof, through data studying the wrong things with increasing precision, through the steady, systematic replacement of everything that matters with everything that can be measured. The joy of discovery — replaced by the performance of proficiency. Curiosity — replaced by compliance. Individuality, difference, wandering, attention, the inner life of the learner — none of it survives contact with a system whose purpose is to sort human beings into an economic hierarchy and credential them for use.
And now the lever is in their hands. Not a reform. Not a policy. Not another rearrangement of the factory floor. The students have it. The same tool — the one that drafts, argues, researches, builds, iterates, and thinks alongside them with more patience than any teacher the system has left standing. Nobody authorized this. Nobody approved it. Nobody convened a subcommittee. They just picked it up. And the system — the factory, the bureaucracy, the hundred-and-fifty-year-old machine — is scrambling to ban it, contain it, regulate it, metabolize it into another compliance exercise before anyone notices that the walls are transparent.
I. The Pattern
Every educational technology follows the same trajectory. A tool that could expand student agency gets adopted by the institution and repurposed to serve the institution’s needs.
The learning management system was supposed to be a learning tool. It became a compliance architecture — assignment tracking, grade recording, attendance logging, liability management. It redefined what counted as a course: content chunked into modules, proficiency demonstrated through submission, learning measured by completion. The tool shaped the pedagogy. The pedagogy never shaped the tool.
The rubric was supposed to create transparency. It became a mechanism for transferring evaluative authority from the student’s own judgment to the system’s predetermined categories. A rubric does not teach you to know when work is good. It teaches you to perform against someone else’s definition of good. The cognitive work of evaluation — the most important cognitive work there is — gets offloaded upward, into the institution, away from the learner.
The standardized test was supposed to ensure accountability. It became the organizing principle of the entire system — curriculum, scheduling, funding, teacher evaluation, school ratings, real estate values. A tool designed to measure learning became the thing that determined what learning meant.
The online course was supposed to democratize access. It became a monetization engine. The xMOOC promised universal education; it delivered universal content and a 4% completion rate. The product was never learning. The product was the appearance of access, sold at scale.
This is not a series of accidents. It is a law. Every tool gets bent toward institutional legibility, compliance, and control. Not through conspiracy. Not through malice. Through the same force that bends rivers toward the sea: systems optimize for their own persistence.
I have watched this happen from the inside for twenty-five years. I have designed learning systems that embedded real pedagogy into the tool itself — visual thinking, collaboration, the portfolio as primary artifact rather than the grade as terminal output. They worked. The system would not adopt them. I have designed assessments that tried to measure what actually matters — the quality of thinking, the depth of iteration, the capacity to respond to critique. The system wanted rubrics. The system wanted data. The system wanted legibility.
The system got what the system wanted.
II. The Break
AI is structurally different from every previous educational technology.
The LMS was a force multiplier for institutions. The rubric was a force multiplier for evaluators. The textbook was a force multiplier for curricula. Every prior tool amplified the system’s capacity to do what it was already doing — deliver content, measure compliance, sort students.
AI is the first tool that multiplies for everyone simultaneously. The student, the teacher, and the institution now hold the same lever.
This has never happened before.
A student with a capable AI has, for the first time, access to something that amplifies their thinking rather than measuring it. They can draft, iterate, argue, research, build — at a pace and scale that was previously gated by access to expert human attention. The asymmetry that the entire system was built on — the institution knows things the student doesn’t, and controls access to evaluation — has dissolved. Not weakened. Dissolved.
The system’s response tells you everything about its actual priorities. Detection. Prohibition. Controlled adoption dressed up as “AI literacy.” These are not educational responses. They are power responses. The predictable moves of an institution whose authority rested on information asymmetry, confronting a tool that has eliminated it.
Meanwhile, in Cambridge, Massachusetts — one of the most educated zip codes on the planet, home to Harvard and MIT, spending forty thousand dollars per student per year — the school district has been trying to get Algebra I into the middle school curriculum since 1992.
Thirty-three years.
The School Committee has passed motions, hired math directors, adopted curricula, fired a superintendent, and fought multiple election cycles over this single question. They eliminated an accelerated program that was working because it was racially inequitable — and they were right to see the inequity. But they never built the replacement. They just removed the thing that worked for some kids without building the thing that would work for all kids. For years, the interim solution was a program the district’s own website described as one that “does not provide explicit in-person instruction of Algebra 1 concepts.”
The district’s algebra program did not teach algebra.
Now, finally, there is an accelerated program. A sixth-grader in the program described her experience — corroborated by her teachers:
“The teacher might give you a baseline idea... like five minutes. And then you do the worksheet on your own and teach yourself the math.”
Five minutes of instruction. Then a worksheet. Thirty-three years of institutional effort produced this. This is our vision for our children’s future.
III. The Machine
The system’s job is to replicate the culture that built it. When the culture values sorting, the system sorts. When the culture values compliance, the system produces compliance. This is what education has always done — transmit the operating assumptions of the civilization into the next generation. When the civilization is healthy, this is how knowledge survives. When the civilization is sick, this is how the sickness spreads.
The teachers are trained inside the machine. The administrators are produced by it. The ones who fight it — who maintain independent judgment, who try something new, who refuse to reduce a child to a data point — do not last. The system does not fire them. It exhausts them. It buries them in documentation. It questions their rigor. It promotes the ones who comply and loses the ones who think. And then it celebrates the rare survivor — the teacher who changed everything, the one who broke through against all odds — as proof that the system works. Why make the exception the hero when you could build a system that doesn’t require heroism?
If you hand AI to teachers trained to deliver predetermined outcomes inside a compliance architecture, you get predetermined outcomes delivered faster. If you hand AI to students inside a system that has never asked them to exercise judgment, you get students who use it to avoid exercising judgment. The tool is not the problem. The tool has never been the problem.
The problem is that we have built a system that does not want what it claims to want. It claims to want thinking. It produces compliance. It claims to want equity. It produces sorting. It claims to want innovation. It metabolizes every innovation into activity — converts structural change into procedural complexity, produces the appearance of reform while deepening the architecture of control.
The system is injust. It sorts children by property value and zip code and names it merit. It assigns futures by policy and names it data.
It is inefficient. Thirty-three years and forty thousand dollars per student to not teach algebra.
It is inequitable. It designs programs for the idea of a student — political, not pedagogical; convenient, not comprehensive — and calls this equity. The real child, with her specific gifts and specific needs, is not in the room.
And it is — by its own stated aspirations, measured against its own declared purpose — a failure.
It does not produce thinkers.
It does not produce citizens.
It does not produce whole human beings equipped for a life of the mind and heart.
It enforces compliance, and credential, and a workforce trained to sit still. And it calls this education.
AI has not broken this machine. AI has revealed what the machine is. It has shown us — by being the first tool the machine cannot fully absorb — exactly what the machine’s actual priorities are. And those priorities are not learning.
IV. The Lever
I spent fifteen years trying to fix this from the inside. I built the alternative tools. I designed the alternative assessments. I trained the teachers. I worked in wealthy innovation schools and in Title I nonprofits and in every configuration between. I wrote the letters and sat in the meetings and watched the system nod politely and change nothing.
The system cannot be reformed from the inside. Not because the people inside it are bad. Because the architecture is self-reinforcing — the training produces the teachers who reproduce the training, the assessment produces the curriculum that produces the assessment, the tools produce the pedagogy that demands the tools. It is, to borrow from Heller, some catch.
But the lever is not waiting for reform. The lever exists. Every student with access to a capable AI now holds a tool that amplifies thinking rather than measuring it — and the system never developed the capacity to evaluate thinking, only output. The asymmetry is already gone. The institution’s walls are not falling. They are being revealed as walls, where before they were called rooms.
What comes next will not be authorized. It will not be approved by a subcommittee. It will not arrive as a policy update or a professional development workshop or a new line item in the technology budget. It will be built by the people the system underestimated — the students who pick up the lever, the teachers who refuse to stop thinking, the practitioners who keep building even when the institution will not adopt what they make.
The system’s production of compliance is itself the destruction of agency. It trained students not to think, and now a tool has arrived that rewards thinking. The lever is also a weapon. Most of the students holding it do not yet know what they have. But the ones who figure it out — the ones of indomitable spirit, the ones the system could not break — will use it not just to learn but to subvert, to route around the system entirely.
Not by protesting it. Not by reforming it. By making it irrelevant.




What a brilliant essay. The writer has given us the picture of the education process that we all should have seen, having gone through it ourselves and having sent our children through it. The effects are seen everyday in the emptiness of personality in so many of our young people and even our peers. The writer makes a beautiful case for AI and how we can use it for good.
Interesting... where is the evidence that, for students, AI "amplifies thinking"?
There seems to be more of an indication that it provides an opportunity for cognitive offloading.