What The Guide Cannot Touch

On August 12, an AI-first private school opens two new campuses in Oklahoma, one in Edmond and one in Tulsa, at forty thousand dollars a year.¹ As of the last public count on July 20, twenty six students were enrolled at the Edmond campus and thirty two at the Tulsa one.¹ The morning at each school is two hours of adaptive software. The rest of the day is workshops.¹ What replaces the teacher in the room is not, technically, a teacher.

Alpha School calls the adult in the room a guide. Guides earn between one hundred and one hundred fifty thousand dollars a year. Guides are not required to hold a teaching certificate. And, by the school's own design, guides do not touch the curriculum.¹

The line to read twice is the third one.

The Choice Under The Word

Every private school this century has tried out a soft synonym for teacher. There are coaches and mentors and facilitators. The choice by itself is not new. What is new is what Alpha's choice pairs it with. The two hours of academic instruction in the morning are not a scaffold the guide fills in. They are delivered, from beginning to end, by adaptive software. The guide is not there to teach the material. The guide is there to keep the room moving through it.¹

MacKenzie Price, the school's co-founder, has said the same thing plainly for two years now. Her students, she says, learn twice as fast in two hours than they would in a full day at a conventional school.² The rest of the day, freed up, becomes what Alpha calls life skills. Public speaking. Entrepreneurship. Building.² The pitch to a parent writing the tuition check is elegant. Two hours of core academics, at double speed, taught by software that never gets tired.

The Claim The Researchers Cannot Verify

The claim is easier to write on a website than to defend in a peer-reviewed journal. Charles Logan, a researcher at Northwestern's Center for Responsible Technology, has said the underlying evidence for the twice-as-fast claim remains difficult to evaluate because Alpha has not opened its data, methods, or student selection to independent scrutiny.³ The internal report Alpha does publish, based on the MAP standardized test, shows its students outperforming national peers.³ Whether the software is the cause, or the fact of a self-selected forty-thousand-dollar-tuition cohort is the cause, is a question the report cannot itself answer.

None of this makes Alpha a fraud. It makes Alpha a bet. The bet is that a child aged five through fifteen can, in two hours a day, learn what a child at a conventional school learns in a full one, if the tool is sharp enough and the guide is committed enough to the room the tool has arranged.

What The Guide Is Not Allowed To Do

Here is where the design detail lands. If the guide does not touch the curriculum, then the guide is not, at the deepest level, in a position to know what a given student thought during those two hours. The record of the child's own thinking, the pause on the fourth question, the mistake reasoned through and corrected, the moment the child guessed and moved on, sits inside the software. What the guide sees is a dashboard. Green squares. Percentages. Speed.

The teacher at the Edmond public school seven miles away will not have a dashboard that clean. She will have a room of twenty eight children she has watched all week. She will know which one hesitated at long division on Wednesday and which one, on Friday, quietly did it in her head. That layer of knowledge is not on any dashboard. It is in the teacher, and it will not appear anywhere else.

Alpha's model asks a specific and honest question. Can the software carry enough of that layer that a certified adult holding it is no longer necessary? The answer is not yet known. What is worth pausing on is what the model, if it succeeds by its own metrics, quietly loses. The child's thinking becomes legible to the machine and, at the same time, less legible to any adult who is not the machine.

The Four Hours That Are Not On The Dashboard

There is a generous reading of Alpha's day. In two hours the child gets through more content. Then the child gets four hours of what schools have always struggled to make room for. Debate. Building. Talking to another person about an idea. If the trade is real, and the two hours actually teach what a full day used to, it is not a small offer.

There is also a quieter reading. In two hours, the child does what the software can measure. In the other four, the child does what neither the software nor the guide is set up to measure. Nowhere in the day is there an adult who has read the child's paragraph in the morning and can, in the afternoon, ask her to explain how she got there.

At Koan we work on the layer of the day the software does not naturally surface. The pause before a sentence. The paragraph rewritten in different words. The moment the student picked one framing over another and can, later, explain why. That layer only becomes visible when the adult in the room is expected, and equipped, to look at it. Alpha's model is a working answer to what a school looks like when the adult is no longer expected to. The answer is smaller than it sounds.

If your child's academic day fits inside a dashboard, and the adult in the room is not the one reading the paragraph, who is holding the record of what she actually thought this morning?