Why the First Page of Your AI Policy Might Be the Only One That Matters

Earlier this year I had the privilege of interviewing some of the leading educators, researchers and builders working at the intersection of AI and education. These blogs are my attempt to do justice to those conversations, to pull out the ideas that matter most and make them useful for everyone working in schools right now.

Most schools now have an AI policy. Very few schools have teachers who have read it.

This is not a criticism of teachers. It is a description of a structural problem that has existed in education long before AI arrived, and one that AI has made more visible and more urgent than ever before.

The question worth asking is not how we write better policies. It is why policies fail in the first place, and what we can actually do about it.

The problem is not the people

At AIDUCATION26 in Bucharest, I spoke with Sacha van Straten, who holds strategic responsibility for both primary and secondary phases at St Catherine's British School in Athens, and Daire Maria Ni Uanachain, a Learning Programme Lead specialising in instructional design, AI-enhanced learning, and digital policy. Both have spent considerable time trying to make AI policy actually land in classrooms. Their diagnosis of why policies fail was the clearest I have heard.

The teacher, they explained, probably does not really understand the policy because it is written in policy language. Or they were handed a forty, fifty, or sixty page document and told to apply it, without being given the time to do the lesson redesigning that genuine implementation requires. And so they tick the box that says they have read it, and then they go back to teaching the way they have always taught.

This is not laziness or resistance. It is a rational response to an impossible situation. A teacher in a busy school, managing thirty students, preparing for exams, supporting pastoral needs, and navigating an already crowded day, does not have the cognitive bandwidth to absorb and implement a complex policy document on top of everything else.

The research backs this up. An analysis of over four hundred education policies across more than forty education systems, conducted by the Inter-American Development Bank, found that fewer than half showed evidence of meaningful progress or impact. A key factor identified across failing policies was not poor design, but poor implementation. The document existed. The change did not.

Why our brains are working against us

There is a well-established body of research in cognitive psychology that helps explain why this keeps happening. Cognitive load theory, developed by educational psychologist John Sweller, holds that human working memory is limited. When the amount of information we are asked to process exceeds that capacity, our performance and decision-making deteriorate. We do not make better decisions with more information beyond a certain point. We make worse ones.

A comprehensive review published in Frontiers in Psychology found that information overload consistently leads to poor decision-making, decreased productivity, and increased cognitive pressure. Psychologist Davy Lewis described the resulting condition as information fatigue syndrome, characterised by anxiety, difficulty retaining information, shorter attention spans, and lower job satisfaction.

When a teacher is handed a sixty-page AI policy, information fatigue is not a risk. It is a near certainty. The document does not get read deeply. It gets skimmed for the parts that seem most immediately relevant, filed away, and quietly forgotten. This is not a failure of professionalism. It is how human cognition works under pressure.

There is also a deeper problem. Research into why teachers do not implement recommended practices points consistently to a factor beyond cognitive load: ownership. Teachers who had no role in shaping a policy feel no particular stake in its success. They comply at the surface level, or they find ways to quietly work around it. Organisational theorists call this loose coupling, the subtle art of pretending to follow a policy without actually changing what you do. A teacher who places a document on the wall of their classroom but does not change their practice is a classic example. The policy is technically present. The practice is unchanged.

The AI policy problem is not new, but it is newly urgent

This pattern predates AI by decades. But AI has made it more urgent for two reasons.

First, the pace of change means that a policy written in September may be partially obsolete by January. The tools move faster than any document can keep up with. A policy that tries to enumerate every permitted and prohibited AI tool is a policy that is already out of date before it is printed.

Second, the stakes are higher. AI touches academic integrity, student data, safeguarding, teacher workload, and parental trust simultaneously. A policy that is not understood and not implemented does not just fail to improve practice. It leaves a school exposed.

According to research from Jisc published in 2024, clear institutional AI policies do reduce staff anxiety and increase appropriate adoption. The problem is not that policies are useless. The problem is that most policies are not designed to be used by the people who need to use them.

What Sacha van Straten did differently

Sacha van Straten's solution was not to write a better policy in the traditional sense. It was to rethink what the entry point into the policy should be.

He took a long, complex AI policy document and reduced it from sixty pages to six or seven, with the entry point distilled to a single page. Not a summary. Not an abridged version. A genuinely usable first page that answered the only two questions most teachers actually need answered when they encounter a new policy: what do we do, and what do we not do, and why.

That first page is an infographic. It is visual rather than textual. It is specific enough to be actionable but simple enough to be absorbed in under two minutes. If a teacher reads nothing else, they walk away understanding the non-negotiables. The full document is still there for anyone who wants the depth. But the entry point no longer requires a sixty-page commitment before anyone gets any value. Daire Maria Ni Uanachain has independently developed a similar approach in her own setting, which suggests this is not a one-off insight but a pattern that works.

This is not dumbing down. It is intelligent design. It is the application of what we know about human cognition to the way we present information to the people who need to act on it.

What productivity research tells us about this

The principle Sacha van Straten is working from has strong support outside of education. In productivity and organisational psychology, the concept of a minimum effective dose refers to the smallest input required to produce the desired output. Applied to policy, this means asking: what is the minimum a teacher needs to understand in order to behave differently? That is the thing that needs to be on page one. Everything else is supporting detail.

Research into habit formation, developed extensively by James Clear in Atomic Habits and grounded in psychological literature on behaviour change, consistently shows that complexity is the enemy of adoption. The more steps required before a new behaviour becomes possible, the less likely it is to become habitual. Reducing friction at the point of entry is not a shortcut. It is the mechanism by which change actually happens.

A sixty-page policy has infinite friction at the point of entry. A one-page infographic has almost none. The question of which one produces changed behaviour in the classroom is not a difficult one to answer.

The teacher who was never part of the conversation

Both Sacha van Straten and Daire Maria Ni Uanachain raised another point that deserves more attention than it typically gets. Teachers are often given policies to implement that they had no part in creating. The policy arrives from leadership, from the trust, from the local authority, or in some cases from government guidance. The teacher's job is to absorb it and apply it.

This is a structural problem because ownership and implementation are not separable. It connects to a broader question about whether teacher AI competency or student AI literacy should come first — teachers cannot lead what they have not shaped. Research into teacher behaviour change consistently finds that teachers who feel involved in the design of a policy or practice are significantly more likely to implement it faithfully. Teachers who feel it was handed down to them are significantly less likely to.

This is not irrational. It is human. If I have contributed to a decision, I understand the reasoning behind it, I feel some stake in its success, and I am more likely to act in accordance with it. If I was not consulted, I have less context, less ownership, and less motivation to do anything beyond minimal compliance.

The implication for AI policy is that the process matters as much as the document. A policy that was developed with teacher input, piloted in classrooms, and refined based on what practitioners actually encountered will outperform a more comprehensive policy that was written by a leadership team and handed down. Every time.

What this means in practice

None of this is an argument for having no policy. A clear, well-communicated AI policy is essential, particularly as the regulatory environment around AI in education continues to develop. In the UK, the Department for Education's 2024 guidance on generative AI provides a framework that schools need to engage with seriously. Across Europe, the EU AI Act introduces requirements that will affect how schools use and procure AI tools. An analysis of what 33 international frameworks say about AI in schools shows just how rapidly this landscape is evolving. The need for policy is not in question.

What is in question is how that policy reaches the people who need to act on it.

The practical implications are straightforward. Start with the infographic. Write the one page that answers what we do, what we do not do, and why, in plain language, before you write the rest. Involve teachers in the process of drafting it, even at the level of reviewing a draft and flagging what is unclear. Build in time for teachers to engage with it properly, not as an announcement in a staff meeting, but as a structured conversation that gives people space to ask questions and raise concerns.

And then accept that even the best policy will need revisiting regularly, because AI is not standing still and neither are the contexts in which teachers are using it.

The page that actually gets read

The test of any policy is not whether it exists. It is whether it changes what people do.

Sacha van Straten's insight, that the first page might be the only page that actually matters in terms of behaviour change, is one I have been thinking about since our conversation. It is a useful provocation for any school leader currently sitting on a policy document that is technically complete but practically invisible.

The question is not whether your AI policy is comprehensive. It is whether the person standing in front of thirty students on a Monday morning knows what to do because of it. If the answer to that question is uncertain, the first page might be the place to start. And if you want a structured way to assess where your school stands more broadly, this AI readiness framework is a good next step.

This blog is part of a series drawing on conversations from AIDUCATION26, a conference dedicated to AI in education held in Bucharest. If you want to understand where your school stands on AI readiness, the DEEP Education Network AI Literacy Audit is a good place to start: audit.deepeducationnetwork.com