Schools are making AI decisions by accident

A few months ago I was running a teacher development session at a school in Mumbai. Midway through, one of the teachers asked me a question I wasn't expecting.

"Are we even allowed to use AI for lesson planning?"

She wasn't asking me for permission. She was asking because nobody at her school had told her either way. Her principal hadn't addressed it. The academic head hadn't mentioned it. There was no AI policy for teachers, no guideline, no conversation. Just silence — which she had interpreted as a prohibition.

Three teachers in the same room had interpreted that same silence differently. One was using ChatGPT daily to draft worksheets. One had tried it once, felt guilty, and stopped. One hadn't touched it because she assumed it was somehow against the rules.

Same school. Same silence. Three completely different behaviours.

This is what happens when schools have no AI policy. Teachers don't wait. They make their own decisions — quietly, individually, and often anxiously. The school doesn't get consistency. It gets a patchwork.

I've seen this across enough schools now to know it isn't a Mumbai problem or an Indian education problem. It's what happens everywhere when institutions move slower than the tools they're supposed to be governing.

The problem isn't that teachers are using AI in the classroom. The problem is that they're doing it without a shared framework — and in that vacuum, the most cautious teachers hold back while the least cautious teachers run ahead. Neither group is wrong, exactly. Both are just filling in the blanks the school left empty.

A school AI policy doesn't have to be long. It doesn't have to be restrictive. It doesn't have to anticipate every scenario AI might eventually create. It has to do one thing: give teachers a clear position to work from.

What is AI for in this school? What is it not for? What do we expect teachers to disclose to students? What counts as academic integrity in a classroom where AI is present?

Four questions. That's the beginning of an AI policy for schools.

What I've found is that the schools most resistant to writing one are usually the schools that most need it. The hesitation is understandable. Writing an AI policy feels like taking a stance on something that's still changing, and nobody wants to publish a document that's outdated in six months. So they wait. And while they wait, the decisions are being made anyway — just informally, inconsistently, and without leadership's fingerprints on them.

There's a version of this that plays out with students too. Schools that have no stated position on student AI use are effectively leaving that decision to each individual teacher, which means students in different classrooms at the same school are operating under different rules. One teacher accepts AI-assisted drafts. Another gives a zero for using it. The student navigating both has no idea which version of the rules applies on any given day.

That's not a student discipline problem. It's a school governance problem.

The teachers I work with are not afraid of AI for teaching. Most of them are genuinely curious about it. What they want is clarity — not because they can't think for themselves, but because clarity lets them move forward without second-guessing every decision. A teacher who knows her school's position on AI can build a workflow around it. A teacher who doesn't know is just guessing.

Teacher workflow and AI is where this gets practical. When a school has a clear position, teachers can actually design how AI fits into their preparation, feedback, and planning cycles. Without that position, every teacher is running a separate experiment — which means the school is learning nothing collectively.

The conversation schools need to have about AI in education isn't primarily a technology conversation. It's a values conversation. What do we believe learning is for? What role does a teacher play that AI can't replace? What do we owe students in terms of preparation for a world where AI is present in every professional context they will ever enter?

Those questions don't have easy answers. But they have better answers when a leadership team works through them together — with teachers in the room, not just management.

A school AI policy built that way has a different quality than one handed down from above. Teachers who were part of the conversation are more likely to follow it, adapt it honestly when it doesn't quite fit a situation, and flag it when something has changed and the guidelines need updating.

The school that asked me to run a full-day AI session recently didn't come to me because they wanted someone to tell them what to think. They came because they wanted a structured way to think through it together. The morning was for teachers. The afternoon was for leadership. By the end of the day they had a first draft of an AI policy — not a finished document, but a live one. Something they built in the room, not something I handed them.

That's the conversation worth having. Not "what can AI do in the classroom" — that changes every few months. But "what do we believe, and how does that shape the decisions we make about AI in this school."

The schools that have that conversation now will be better placed for the next tool, and the one after that. The schools still waiting for the dust to settle are going to be making AI decisions by accident for a long time.