The Largest School System In America Says Wait

Early in July, Kamar Samuels, the chancellor of the New York City public schools, sent a note to his principals with a short and unusual request. Hold off on buying new educational software. Software the district needed to open a building could still be purchased. Everything else, for now, was on pause, until the department finished the AI guidance it had promised in June and, then, quietly deferred.¹

The system Samuels leads is the largest in the country. Close to a million children will walk through its doors in a few weeks. Less than a month before they do, the guidance is still not out.²

What The Wait Means

The wait did not begin as the chancellor's idea. In June a bipartisan majority of the New York City Council sent a letter to Mayor Zohran Mamdani and to Chancellor Samuels asking for a two-year moratorium on the use of AI in the schools. The council members cited concerns about learning and about mental health. Nearly six thousand five hundred comments had been filed on the March draft of the district's AI policy. Parents, teachers, and, in a separate letter, hundreds of artists asked for the guidance to be stronger, slower, or absent altogether.³²

What Samuels offered in reply was not a moratorium. It was a hold. The distinction matters. A moratorium is a decision. A hold is a pause. A hold is what a system does when it has run out of vocabulary and has decided not to buy anything until it has some. It is the least glamorous move a large organization can make. It is also, this summer, one of the more honest ones.

The Sound Of A System Thinking

Every fall for the last five years the sound has been the same. A district signs. A vendor ships. A pilot begins. A dashboard lights up in the principal's office. By the spring the pilot is a program and the program is a line in the budget. The next fall the line is doubled, and by the spring after that the district cannot say, in a single sentence, what any of it changed. The wagons empty out and the district looks up and asks what it bought.

What the New York pause is asking, in effect, is for the wagons to slow down. Not because the tools are worthless. Because the district cannot yet describe the thing it wants the tools to leave behind. And a district that cannot describe what it wants cannot procure it. It can only procure what the vendors have already named. In 2026 what the vendors have named is a chatbot, an assistant, a study buddy, a lesson planner. Useful nouns. All of them describe outputs. None of them describes the record of a student's thinking after a semester with the tool.

The Column Nobody Has Filled In

Chancellor Samuels' office is reported to be working on a scorecard for teacher-facing AI products, and on a more rigorous review process before new tools can be adopted.¹ A scorecard is a discipline. It is a set of columns a district has decided to fill in before it signs. Ease of use, of course. Bias, of course. Data safeguards, of course. The harder column, and the one nobody has written well yet, is the one that asks what the tool leaves behind in the student.

A demo shows a chatbot answering a proof. A semester with the tool will show something else. Whether the child, at the end of it, can walk her teacher through the proof unaided. Whether the six revisions she made to the same paragraph in September were her own. Whether the twenty-minute pause on last Tuesday's essay was the pause of a child thinking or the pause of a child waiting for the machine.

What A Hold Buys

At Koan we spend our days on the second question. The layer we build sits inside the writing environment the child is already in and quietly leaves a record of the work as it happened. The pause. The revision. The passage typed by hand and, later, the paragraph pasted in from another window, and then, sometimes, rewritten. Not a probability score. A record.

The reason a record matters to a scorecard is that a scorecard is only as good as the evidence a district can point to when it fills it in. If New York, at close to a million children, asks a vendor what the tool left behind in the student, the vendor will need to be able to show more than daily active use. The industry has spent five years selling engagement. The next five will be about showing what the engagement produced.

The Kind Of Slowness Worth Buying

The New York pause is a small act, on paper. In effect, it is one of the more consequential moves any large district has made this year. Not because a pause solves anything. Because a district that admits it does not yet know what it is buying is a district that is, at last, taking the question seriously. The vendors who will be there when the pause lifts will be the ones who can describe, in a language a scorecard can use, what their tool leaves behind in a child at the end of a term.

For the rest of the country, watching from a distance, the question is not whether to copy the New York hold. Most districts do not have the political weight to pause a market on their own. The question is what to do with the hold's premise, which is that a purchase order without a description of what a tool should leave behind is a purchase order for a demo. A demo, three years in, is not the thing a district can point to when a parent, or a state, or a student herself asks what the fall was for.

If your district paused its purchases tomorrow, what would you write in the column that asks what the tool leaves behind in the student?