I want to start by saying something most of you already know but rarely hear out loud. The conversation we've been having for the last twelve months — the one about whether to allow AI in our schools — was always going to age badly. It has aged badly. AI is in our students' homework. It is in our colleagues' planning. It is in the writing on this slide deck. The question of permission was answered by the technology, not by us.
The question we should have been asking, and the question this talk is about, is fluency.
Fluency is not a policy. Fluency is what your faculty does on a Tuesday morning, with no one watching, when they sit down to plan a lesson. The fluent teacher reaches for the tool the way they reach for a textbook — habitually, without ceremony, with a clear sense of when it helps and when it doesn't. The non-fluent teacher reaches for the tool nervously, or doesn't reach at all, and the gap between those two teachers compounds week over week. By the end of a school year, those two teachers are not in the same profession.
That is the imperative. Not whether to allow it. Not how to police it. Whether the people you employ to teach can teach with this thing in the room.
So let me get specific about what fluent looks like. A fluent teacher uses AI to differentiate a lesson three ways before they walk into the classroom. A fluent teacher uses AI to draft the parent email they were dreading writing, then edits it for tone in the time it would have taken them to write a first sentence. A fluent teacher asks AI to play the role of a struggling student and tries out three different ways of explaining photosynthesis until one of them lands. A fluent teacher does these things in twenty minutes a day, every day, and at the end of a term they have produced a different kind of classroom from the one they had in September.
A non-fluent teacher does none of these things. Not because they can't. Because no one has shown them that twenty minutes a day for six weeks is what changes them.
Which brings me to what leadership actually has to do here. You don't write a policy first. You build practice first. You take a cohort of fifteen teachers — pick the curious ones, not the senior ones — and you give them six weeks of structured time to use AI on their actual work. Real lessons. Real assessments. Real parent communications. You meet weekly. You share what worked. You share what didn't. At the end of six weeks, you have fifteen teachers who can describe AI fluency from the inside, in their own classrooms, in their own subject areas. Those fifteen teachers become your faculty's centre of gravity. They are who write your policy. Not your inspectorate. Not your board. Your fluent teachers.
The schools I'm seeing succeed at this are running this loop right now. They are not waiting. They are not asking permission. They are not commissioning external reports. They are putting fifteen teachers in a room every Wednesday afternoon and watching the room change.
If you take one thing from this talk, take this. Teachers aren't being replaced by AI. They're being replaced by teachers who use it. Your job, as a leader, is to make sure the teachers in your school are in the second group. You have less time than you think to make that happen, and the only way to do it is to start.
Thank you.