Of the five commitments named in the ministerial statement adopted at Digital Learning Week 2026 — public accountability, deliberative governance, teacher agency, learner rights, and safety — teacher agency is the one that speaks most directly to a worry most parents already have, even if they've never used the phrase: is a real person still in charge here, or is the software quietly running the show?

What UNESCO actually committed to

On 9 September 2026, more than 25 education ministers and government representatives adopted a joint statement arguing that education should "remain a common good in the age of AI." Teacher agency was named explicitly as one of the things AI's arrival in schools must not be allowed to erode.

The concept is narrower and more practical than it sounds. It isn't a vague statement that "teachers matter" — it's a specific claim about decision rights. A teacher's professional judgement — what a student needs next, whether they've genuinely understood something, how to respond to a struggling class — should stay a decision a human teacher makes, with AI tools supporting that judgement, not substituting for it or making it invisibly on the teacher's behalf.

Why this needed saying out loud

It's easy to assume no reasonable AI product would ever try to replace a teacher's judgement outright — and mostly, none do, at least not on purpose. The real risk UNESCO is naming is quieter than that. AI tools are increasingly capable of doing things that used to require a teacher's active decision: grading a piece of work, flagging which student needs extra support, suggesting what topic to cover next based on a data pattern rather than a teacher's read of the room.

None of those capabilities are inherently bad. The problem is how easily the handover of that decision can happen without anyone quite noticing it did. A school adopts a tool because it saves time on grading. Six months later, the tool's suggestions are quietly shaping what gets taught, because following them is easier than overriding them, and nobody made an explicit decision that the AI should have that much influence — it just accumulated.

Teacher agency, as a protected principle, is UNESCO's way of saying: that accumulation needs to be a deliberate choice a school makes and can see clearly, not something that happens by default because a tool is convenient.

The version of this that shows up at home

Most parents reading this don't run a school — but the same underlying principle shows up, in miniature, in any AI tool your child uses without a teacher in the room at all. Swap "teacher" for "you, or your child's own judgement" and the question is identical: is the AI supporting a decision that should stay human, or has it quietly started making that decision itself?

Concretely, that shows up in things like: does the tool decide what your child studies next, or does it help your child (and you) decide? Does it tell your child whether they've understood something, or does it ask questions that let your child discover whether they've understood it? Does it decide when your child has done enough for the day, or does it hand that judgement back to a tired, honest human?

A Socratic approach to AI tutoring — one built around asking rather than telling, hinting rather than answering — is, in effect, teacher agency implemented at the level of a single child and a single session: the tool's job is to support the thinking, not replace the judgement about what the thinking should conclude.

Why the distinction is easy to miss

The tricky part is that a tool eroding agency and a tool genuinely helping can look identical from the outside for weeks. Both save time. Both feel efficient. The difference only shows up over a longer horizon: does the human — teacher or parent — still know why a decision was made, and could they have made a different one if they'd disagreed? Or has "what the AI suggested" quietly become the only option anyone actually considers?

That's the test worth applying, at school-policy scale or at your own kitchen table: not "does this tool help," almost everything marketed as an education product claims that, but "who's still actually deciding, once the tool is in the room."

FAQ

What does 'teacher agency' mean in the context of AI?

It means a teacher's professional judgement stays genuinely in charge of what happens in a classroom, even when AI tools are involved — the AI supports the teacher's decisions rather than making them invisibly in the background or nudging the teacher toward a particular outcome without their awareness.

Why did UNESCO's ministers specifically commit to protecting teacher agency?

Because AI systems are increasingly capable of doing things that used to require a teacher's judgement — grading, flagging a struggling student, suggesting what to teach next — and it's possible for a school to adopt those tools in a way that quietly hollows out the teacher's actual decision-making role rather than supporting it.

Does teacher agency matter for an AI tutor my child uses at home, with no teacher involved?

Yes, in a related form. At home, the equivalent question is whether the tool defers to your judgement and your child's own effort, or quietly takes over decisions that should stay with a person — what to study next, whether an answer is understood, when to rest. The same principle, no classroom required.

Duke Harewood built aitutors.me around a Socratic model precisely so the judgement about a child's understanding never quietly transfers to the software — it stays with the child, checked by questions, not answers. Published 11 September 2026.