Introduction

Introduction

Who this book is for

This book is for people who make or influence decisions about AI in education, and for anyone who wants to understand those decisions. That includes teachers, school leaders, parents, governors, policy makers and the people who sell to them. It assumes no technical background. The chapters are short on purpose, so you can read one on a train, or pick the part that matches the problem on your desk.

What it argues

Three ideas run through every part.

The first is that choices matter more than capability. AI will keep improving, but whether that helps or harms learning depends on decisions about purpose, pace, safeguards, money and who is included.

The second is that evidence comes in different strengths, and a good decision says which kind it relies on. A single study, an expert survey, a vendor's claim and a long-running trend are not the same thing. Part One shows one way of labelling that difference, and every part ends by saying what it can't tell you.

The third is that human relationships are how people learn. Tools can help or get in the way. Where a chapter has to choose between a more efficient tool and a protected human relationship, it leans towards protecting the relationship, and it says why.

How the book is organised

There are eleven parts and 56 short chapters. Each part is built from one research paper that I wrote from my newsletter, and can be read on its own.

Part One looks ahead to 2035 and 2045 through scenarios, and names the choices that decide which future arrives.

Part Two asks how people learn with AI, and tests an expert consensus against the empirical evidence.

Part Three turns to children and young people, and to wellbeing, attachment and the developing mind.

Part Four compares international definitions of AI literacy.

Part Five looks at assessment: what a piece of work can still prove when AI can produce it.

Part Six looks at the labour market and at what schools should prepare young people for.

Part Seven puts teachers at the centre, with professional learning, time and the right to say no.

Part Eight is about school leadership and the order in which to make changes.

Part Nine covers governance, rules and procurement.

Part Ten asks whether AI will close or widen the gaps between learners, especially those in crisis.

Part Eleven explains how the papers and this book were made, and what checking found.

Most parts follow the same pattern: an opening note on where the material comes from, short chapters, a section called What to do, a plain statement of what the part can't tell you, and a closing thought. Sources appear as numbered citations in the text, and the full list is in the Bibliography at the back. Studies that matter to more than one part are explained in full once, and later chapters point back to that place.

How to read it

Read it in order if you can. Part One sets the frame and Part Eleven explains the method. If you're short of time, read Part One, then the part closest to your role: Part Seven for teachers, Part Eight for school leaders, Part Nine for governors and policy makers, Part Three for parents. The What to do sections are written to be taken to a meeting.

A note on evidence

I've tried to report numbers as the sources give them, and to say who is making a claim. Claims that organisations make about their own programmes are reported as their claims. Several figures from the original newsletters were corrected or removed when they couldn't be traced to a source. Even so, errors are possible. If a number matters to a decision, please check it against the source cited.

Who wrote it

This book was co-written by me and an AI agent, Claude, made by Anthropic. I chose the themes, wrote the newsletters the papers draw on, and set the standards. Claude gathered and checked sources, synthesised the themes and drafted the text. I reviewed and approved the book, and I'm responsible for it. Part Eleven gives a fuller account, including the errors that checking found.

That arrangement is itself a conflict of interest, because several parts discuss Anthropic and other AI developers. I've also led international schools, and work with Brave Generation Academy, an alternative education provider discussed in Parts Eight and Ten. I flag these interests in the chapters where they matter.

A last word before you begin

The title is a statement of intent. To be deliberately human is to use powerful tools on purpose, with the people we serve kept in view. It's a choice, and it has to be made again each time a new tool, contract or rule comes along. This book tries to make that choice a little easier to see.