Four Ways School Leaders Should Be Using AI Already

Adam Sturdee
May 27
7 min read

Most discussion of AI in education is a classroom discussion. Can teachers plan with it? Can pupils use it safely? Can it cut marking time? Can it generate quizzes or model answers? These are useful questions, but they are not the most important ones.

The more important question is now a leadership one. How should headteachers, deputies, assistant heads, CEOs and directors of education actually be using AI to think, decide, communicate and lead?

How that question is answered will shape almost everything else. Policies, training, procurement, culture, adoption: all of it will be filtered through the quality of leadership judgement around AI. If leaders use AI superficially, their schools will follow. If leaders delegate all AI thinking to the enthusiastic middle leaders, they will misread both the opportunity and the risk. If leaders treat AI as a shortcut, staff will notice.

The schools that make the best use of AI over the next five years will not necessarily be the schools with the most tools. They will be the schools where leaders themselves develop disciplined habits of thinking with AI.

The quiet leadership gap

In conversations with other school and trust leaders, I see three rough patterns.

The first is the well-informed leader. They read widely, attend conferences, can name the latest model and follow the policy debate closely. They know the language. They have not yet built AI into their own working week.

The second is the weekend experimenter. They have generated resources, tested half a dozen tools and shared clever examples in the staffroom. The experimentation has not yet connected to the actual complexity of leadership work.

The third is the strategist on paper. They have written the policy, set up the working group, perhaps drafted a vision document. They still do not personally feel that AI can support the kind of high-stakes work they do.

All three are making progress. All three are leaving most of the value on the table.

Senior leadership is not generic work. It involves judgement, nuance, competing priorities, undocumented context and incomplete information. Asking AI to summarise a policy or draft a routine email saves a few minutes. That is not where the value lies. The deeper value comes when AI is used to clarify thinking, test assumptions, surface blind spots and build better structures for decision-making. That requires more than casual use.

Before output, context

There are dozens of prompting tips circulating in education. Most of them miss the main point.

The single biggest determinant of AI usefulness for a school leader is not the prompt. It is the context the leader provides.

So much of what shapes a leadership decision in schools is undocumented. A staffing issue sits inside years of relationship history. A parental complaint sits inside a wider pattern of community anxiety. A behaviour policy looks reasonable on paper and lands very differently in Year 8 than in Year 11. A trust-wide decision can be technically efficient and culturally clumsy at the same time.

AI cannot know any of this unless the leader provides it.

A useful habit is to begin with a brain dump before asking for anything. Here is the situation. Here is what I think is going on. Here is what I am worried about. Here is what I am trying to avoid. Here is what I need to decide. The notes do not need to be polished. They often work better when they are not.

A second habit is more powerful, and it surprises leaders the first time they try it. Before asking AI for a recommendation, ask it to interview you. “Before you advise me on this, ask the questions you need answered to understand it properly.” “Challenge my assumptions before helping me decide.”

The reason this matters is that in school leadership the first version of a problem is rarely the real problem. A request for help with a staff communication is often about trust. A workload concern is often about implementation design. A resistance issue is often about timing, fear or a lack of clarity. AI is genuinely good at probing for those layers, but only if you give it permission to do so before it generates output.

The discipline this builds is not really about AI. It is about reflective leadership. The act of briefing AI properly improves the leader’s own thinking before any answer is generated.

Four functions, not four tools

Once that habit of briefing is in place, the same approach can be applied to four distinct leadership functions. None of them require specialist software. All of them require disciplined use of whichever general-purpose tool is to hand.

1. AI as a research analyst

School leaders work in information-rich, time-poor conditions. Policy documents, inspection updates, safeguarding guidance, attendance data, behaviour patterns, SEND pressures, staff survey results, trust priorities, local context. No senior leader has time to read, compare and synthesise all of it manually.

The mistake is to use AI like a search engine.

A weak prompt: “What should our AI policy say?”

A stronger prompt: “We are a secondary school in England developing an AI policy for staff

and pupils. We need to balance innovation, safeguarding, academic integrity, workload and GDPR. Prioritise UK education sources. Avoid vendor marketing. Distinguish between statutory requirements and recommended practice. Identify the key decisions governors will need to understand before approving it.”

The leader is not asking a search-engine question there. They are briefing an analyst.

The same shift applies to CPD design, attendance interventions, curriculum change, governance reporting, behaviour strategy, inspection preparation and parental engagement.

Two safeguards matter. First, check claims. AI is fluent before it is accurate. Three questions catch most failures. Is this grounded in real sources or is it pattern-matching? What is missing that I did not think to ask? Would I be comfortable putting my name to this? If the answer to the third is no, the work is not finished.

Second, send the same complex research question to more than one model. Where they agree, the answer is probably reliable. Where they diverge, that is exactly where you need to look more closely.

2. AI as a strategic thought partner

Senior leadership can be lonely. Even in healthy teams, there are decisions not yet ready to share. Early thoughts that need testing. Tensions that need careful handling. Governor or trustee conversations that need preparation.

AI cannot replace a trusted colleague. It can, used carefully, create a structured space for challenge.

The most useful prompts are not “what should I do?” They are closer to: “Help me think this through. Act as a careful school leadership adviser. Identify the strongest argument for this decision and the strongest argument against it. Surface the likely staff concerns, the safeguarding implications, the workload implications and the conditions that would need to be in place for it to work.”

This works because it slows the leader down at exactly the moment they would normally rush. It also pushes back against two predictable leadership errors with AI. The first is assuming it is a gimmick because the leader has only used it for basic tasks. The second is assuming it can solve complex organisational problems without culture, training and trust.

Both errors are more likely when leaders do not use AI seriously themselves.

3. AI as a communication partner

Communication is where the gap between weak AI use and strong AI use is most visible. Generic AI prose is now easy to spot in schools. It sounds polished, slightly bland, slightly detached. Staff identify it quickly, and once they do, the leader’s credibility takes a small hit.

The fix is not to avoid AI for communication. It is to train it on voice, audience and purpose.

Two practical moves help. First, feed AI a collection of writing the leader is happy with, and ask it to analyse the style. AI is good at naming patterns leaders cannot articulate themselves. The analysis becomes a working style guide that can be applied to future drafts.

Second, use AI as a stakeholder review panel. Before sending a sensitive message, ask it to read the draft from a particular perspective. Review this as a sceptical teacher who is worried about workload. Review this as a parent who may misunderstand what AI is being used for in school. Review this as a governor concerned about safeguarding. Review this as a union representative looking for any sign of surveillance framing.

This is not manipulation. It is empathy and clarity at scale. The best school communication anticipates concern without becoming defensive, and AI can help test whether a message is clear, proportionate and likely to build trust before it is sent.

4. AI as an operational intelligence layer

Schools generate large amounts of information that never quite becomes insight. Behaviour logs, attendance data, safeguarding themes, CPD feedback, lesson visits, staff surveys, parent communications, improvement priorities. Leaders carry much of the pattern recognition in their heads.

AI can help with synthesis, but this is the area where the governance question matters most.

The purpose must be visibility, not surveillance. A trust leader can use AI to synthesise anonymised themes from school improvement conversations. A headteacher can use it to identify recurring workload concerns from staff feedback. A CPD lead can use it to spot patterns in coaching conversations. A governor report can be sharpened by clearer summaries of evidence, risk and progress.

The discipline here is to start manually before automating anything. Run the process by hand several times. Check whether the insight is genuinely useful. Refine the questions. Involve the right people. Consider the data protection implications. Only then build anything systematic. The question for schools is never simply “can we automate this?” It is “should we, what purpose does it serve, who benefits, what risks does it create, and where must human judgement remain central?”

The leadership question

The real test for AI in schools is not whether it can produce content. It can. The question is whether it can help leaders think more clearly, act more wisely and build schools where technology strengthens rather than weakens professional culture.

Used badly, AI will produce more noise, more documents, more half-finished initiatives, more confusion. Used well, it can protect time, surface insight, improve communication, sharpen strategy and support professional judgement.

That last point matters most. Professional judgement does not become less important in the age of AI. It becomes more important. The leaders who develop disciplined habits of working with AI will not lose their judgement to it. They will sharpen it.

Most of the AI conversation in education is still about tools, prompts and policies. Those things matter. But the more durable shift is quieter. It is leaders themselves moving from passive awareness to disciplined practice, not because AI should replace leadership, but because leadership now has to understand AI.

The four-function framing in this article was prompted by Nufar Gaspar’s discussion of AI for executives on The AI Daily Brief. The translation into school and trust leadership is mine.

Adam Sturdee is a senior leader and co-founder of Starlight, the UK’s teacher-first AI-powered transcript-based coaching platform for educators.

His work sits at the intersection of dialogic practice, instructional leadership and responsible AI strategy for schools and trusts.

He recently presented his research on AI-supported coaching at the BERA TEAN Conference 2026: https://www.bera.ac.uk/conference/bera-tean-conference-2026