Leading a school through the AI question

The two failure modes

The first is prohibition. It is quick, it feels responsible, and it fails — the tools are on everyone’s phone, so the practice continues where the school cannot see it, and the staff most likely to have come and asked are the ones who stop.

The second is enthusiasm without boundaries. A school buys a platform, runs a launch day, and nobody has written down what may never be entered or who checks output before it reaches a parent. This one is more dangerous because it looks like progress.

A position staff can act on

What staff want is not a strategy document. It is to know whether the thing they did last Tuesday was allowed.

• Name what is permitted, in examples rather than principles — planning, differentiation, first drafts, rephrasing.
• Name what is never permitted, as a short list of data categories, not a judgement call.
• Say who to ask when something falls between the two, and make it safe to ask.
• Say what happens when a mistake is made, before one is made.

Say the workload part out loud

The reason this is worth leading rather than suppressing is workload. The administrative weight on teachers and SNAs is a real cause of people leaving, and some of it is genuinely reducible — the drafting, the rephrasing, the summarising, the forms.

That is a legitimate thing for a school leader to want, and saying so is more honest than framing the whole question as risk management. The risk framing on its own tells staff the school would rather they did not, which returns you to failure mode one.