Where The Worry Went

Last Tuesday, June 30, the Innocenti office of UNICEF released a study of children and AI drawn from surveys in ten countries, from Armenia to Brazil to Pakistan.¹ The headline was the ratio. Children are adopting AI systems more than three times faster than their parents and caregivers. The absolute number matters too. At least twenty million children have used one of these tools, and up to half the children surveyed said they were already using them regularly.²

Two numbers inside the study are worth reading slowly. Thirteen million children reported using AI for homework. That number is what most of the coverage led with, and it will show up in the next round of district policy memos. The other number is smaller and stranger. About two million, roughly one in ten of the children surveyed, said they were using AI for advice about things that worried them.³ The work was led by Innocenti with ECPAT International and INTERPOL, funded by Safe Online, and covered roughly a thousand children ages twelve to seventeen per country.⁴

The Room The Adults Are Not In

The three-to-one ratio is easy to gloss past as a curiosity. It is not a curiosity. It is a description of a room. In that room, the child is the fluent one and the adult is not. She is not asking the parent, not asking the teacher, and, in a growing number of cases, not asking the friend either. She is asking the model. And the model is answering, in a room the adults in her life cannot see into, on a device they often did not buy, in a session they will not read.

The homework side of that room is the one the story is going to be about. Districts will read the thirteen million and update their acceptable-use language. A few will run summer training. Some will buy a detector. That work is real. What the story does not yet know how to be about is the other side of the room. The one in ten. The two million small conversations where the question was not "help me finish this essay." The question was closer to "I don't know what to do."

What The Homework Number Is Actually About

The two numbers are not as separate as they look. Any teacher who has spent a semester reading student writing knows the boundary is porous. A worry about a grade becomes a worry about a parent becomes a worry about a future. A homework question becomes a where-do-I-belong question in about three sentences. The child sitting alone with the model at ten in the evening does not carefully sort her prompts into the schoolwork column and the life column. The model does not sort them either. It answers both, in the same voice, in the same session, and the record of that session sits somewhere the child rarely reads back and the adults do not read at all.

What the school has to grapple with in the fall is not whether the homework was done by the child. That framing is losing altitude. It is that the tool is doing more than the homework, that the child knows it is doing more than the homework, and that the record of that fuller work is being made elsewhere. Every time a student sits down with the model, a shadow curriculum is being written. Not by the district, not by the teacher, not by the parent. By the child and the tool, together, in the room the adults are not in.

The Classroom That Could See The Question

The instinct after reading the UNICEF numbers is to want the parents in the room. UNICEF's own framing calls for it, and rightly.⁵ The parent, though, is only one of the seats that has emptied. The teacher's seat has too. She is not in the session where the essay was worked out, and she is not in the session where the essay turned, at ten fifteen at night, into something else.

The classroom cannot follow the child home. It should not want to. What it can do is design its own room so that when the child sits down inside it, the tool she is using leaves a record the teacher can read. Not to police the essay. To read the student. The pause before she typed the prompt. The revision she kept. The place she pushed back on the model, or the place she did not. The moment she asked a question that was not on the assignment. These are the marks the classroom used to be able to see. They are also, on a Tuesday night at ten fifteen, the marks that would tell an adult which room the child was actually in. If a school builds a workspace where those marks are visible again, the homework and the worry stop being two rooms. They become one room, and the teacher, at last, has a chair in it.

Twenty million children have already stepped into the room. One in ten of them, on any given evening, is asking a question that is not about the assignment. The adults are, in nearly every country, still on the far side of the door.

If the record of your students' work with the tool were readable to you tomorrow, what would you want to have seen at ten fifteen last night?