Generative AI in the international school classroom
There was a Sunday evening when I sat down at my desk and stared at a pile of student essays. They needed to be back in students’ hands Monday morning. They were good essays, mostly. My students had worked hard. But I could feel the quality of my attention thinning — moving through the pile rather than into it, writing margin notes that were true but somehow not quite what each student needed to hear. There wasn’t enough of me left by Sunday evening to give the kind of feedback I knew how to give.
I’d been feeling this for a while. The parts of teaching I loved — dreaming up new projects, designing activities that would make students stop and genuinely think, finding the one weird historical connection that could light up a room — were getting squeezed out by the parts that were just administration. The mechanical work of teaching was everywhere that year: essays to mark, lesson plans to template, grammar to correct, report card comments to write. My three sections of AP World History alone generate a constant stream — Document-Based Questions (DBQs), Long Essay Questions (LEQs), Short Answer Questions (SAQs) — and that’s just one course. The machinery of teaching, not the art of it.
I want to be honest about what was underneath that exhaustion. It wasn’t just a heavy workload — it was the particular dread of knowing that what you’re spending your energy on isn’t the part that matters. Teachers leave the profession for a lot of reasons, but one of them, named quietly in exit surveys and loudly in staff room conversations, is that the job as it’s actually lived doesn’t look like the job they signed up for. The creativity, the relationships, the moments when something clicks for a student — those are real, and they’re why most people go into teaching. The Sunday night pile is real too. And for a lot of teachers, the pile is winning.
That’s when I started using ChatGPT. I’d been curious about it for a while, and when I finally sat down with it, the potential was immediately obvious. Not as a shortcut — as a solution to a specific, real problem. Here was a tool that could handle the parts of my job that didn’t require a teacher, and free up the parts that did.
Getting started required the kind of small workarounds that are simply part of teaching in this context: tools that everyone else uses by default sometimes need a subscription, a workaround, or a favor from someone half a world away who just wants to sleep.
At first I used it the way you’d use a slightly over-eager intern: give me five ideas for a classroom debate on this topic. Most of the suggestions were mediocre. Some were surprisingly good. But having a thinking partner — even an artificial one — changed something. I stopped staring at a blank planning document and started reacting, editing, stealing the good parts and throwing out the rest.
Then I realized I could go further. Lesson outlines. Essay feedback on grammar, syntax, spelling — the mechanical stuff that doesn’t require my judgment as a teacher but does require forty minutes of an evening. I handed that over. And when I did, something opened up. The creative work I’d been wanting to do but never quite had the bandwidth for — the interdisciplinary projects, the simulations, the lessons that made students genuinely argue with each other — started actually happening.
The results were not subtle. When I had more time and energy for the creative, relational side of my teaching, students were more engaged. They produced better work. Some of them started to love the subject in a way that hadn’t happened before. I don’t think that’s a coincidence.
What I had more time for — the creative leaps, the relational moments, the lessons that asked students to actually think rather than just produce — turns out to be exactly what AI cannot supply.
There is a version of this argument that gets made in education research, and I find it convincing: the classroom’s most important function was never really knowledge delivery. It was always the relationships — the teacher who knows which student is struggling before the student does, who remembers what a kid said in September when grading their essay in March, who can read a room and decide in the moment whether to push or to ease off. AI can simulate knowledge delivery better than most people expected. It cannot simulate the relationship. It cannot carry the context of a specific child across a full year. It cannot make the judgment call that comes from knowing someone. That’s not a small gap — it’s the whole thing. It’s what the title of this book is about.
Research has been trying to tell us this for years: creativity and emotional intelligence predict success in ways that test scores often don’t, and they’re also the capacities most at risk when any tool starts doing the thinking for us. That tension runs through every chapter of this book.
Nobody has done this before. The teachers working through these questions right now — how to use AI honestly, how to teach students to use it well, how to protect the capacities that make human learning worth doing — are the first generation to figure it out without a roadmap. That is uncomfortable. It is also genuinely interesting work, and I’d rather be doing it than not.
This book is about that opening up. And it’s about doing it carefully and honestly — written from one of the most fascinating and complicated teaching environments in the world, an international school in China, but shaped by questions that reach well beyond it.
Why This Book
The AI-in-education conversation is already crowded. There are books, podcasts, LinkedIn posts, and professional development workshops where someone shows you ChatGPT for twenty minutes and tells you it will change everything. Most of this content is written for an imagined generic teacher in an imagined generic school — usually in the United States or United Kingdom — with full access to every tool available and no particular cultural or regulatory constraints.
That’s not where I teach. And if you picked up this book, it’s probably not where you teach either.
International schools in China sit at one of the most interesting intersections in global education. They operate across a wide range of rigorous curricula: the AP program’s commitment to college-level analytical rigor and exam-based mastery; the IB’s emphasis on inquiry, reflection, and extended independent work; the Cambridge tradition of structured academic discipline; and at schools following American curriculum frameworks, with Common Core standards that mirror the US K–12 framework. What unites all of these is a demanding set of expectations — and a student body shaped by an educational tradition that prizes diligence, deference, and results. They operate in a country where the internet looks different, where many of the tools everyone else is talking about are inaccessible, and where the government has its own ambitious and rapidly evolving vision for what AI in education should look like.
And then there is the pressure. China’s demographic scale creates a level of academic competition that is genuinely difficult to convey to someone who hasn’t seen it up close. For many of the students in Chinese international schools — and for their parents — admission to an elite university, including the Ivy League, is not a vague aspiration. It is a concrete, non-negotiable expectation. The stakes feel existential in a way that shapes everything: how students approach failure, how they relate to their teachers, what risks they’re willing to take, and yes, how they think about AI.
There’s a student I’m thinking of right now. I stopped teaching him two years ago. He still comes back to my classroom.
Not because he has to. Because he wants to. He sits down, asks what I’m working on, tells me what he’s reading, pushes back on things I say. Sometimes he has a question about something from another class and he’s come to me specifically — not just any available adult, me. I think about what that means. Over the course of a year together, he figured out that I would be straight with him, that I’d take his questions seriously, that I’d push back when I thought he was wrong. He keeps coming back for more of that. It’s not complicated, but it’s not accidental either. You have to earn it.
I’ve taught in two very different kinds of international schools. At a boarding school, my students saw their parents on weekends — and those weekends were full of violin lessons and swimming practice and tutoring sessions, so they didn’t really see them then either. At my current school, many of my students come from families wealthy enough that the parents are rarely home in any meaningful sense — traveling for work, managing businesses, living lives that don’t leave much room for a teenager’s ordinary needs. The term for this is affluent abandonment, and it’s more common in international school settings than most school literature acknowledges. The money is present. The parent sometimes isn’t.
What this produces, in my experience, is students who are emotionally hungry in ways that can look like behavior problems or disengagement or attitude — until you realize they’re just looking for an adult who will stay still long enough to actually see them. The troubled kids, the difficult ones, the quiet ones, the ones other teachers find exhausting — those are often the students who end up in my room. I think it’s because I don’t find them exhausting. I find them interesting. And I think, honestly, that being a father helps. I know what a kid who needs something but can’t say so looks like. I’ve seen it at home.
None of this is in any AI tool I’ve used. The software doesn’t know that this student’s mother is in Beijing this month, or that the quiet one in the back row hasn’t eaten breakfast in three days because the housekeeper called in sick, or that the reason the essay came in late isn’t laziness but a 2am phone call home that went badly. It doesn’t know which students need a direct challenge and which ones need someone to sit with them for a minute before the work starts. It doesn’t know, and it can’t know. That’s what this book is actually about.
And your students are already working through all of this themselves. They have access to the same tools you do — more, in many cases, given how quickly teenagers find corners of the internet that adults haven’t reached yet. The question was never really whether AI would enter your classroom. It was whether you’d be part of the conversation about how.
There are four groups of people dealing with AI in schools right now, and they’re each dealing with a different version of the same problem. Students are using it — most of them, most of the time, in ways their teachers may or may not know about. Admin and tech departments are trying to write policy for a tool they’ve often never used in a classroom. Coaches and curriculum leads are supporting teachers through a transition that nobody has a roadmap for. And teachers are doing all of it at once: learning the tool themselves, teaching with it, teaching students about it, and operating inside frameworks and policies they had little input on — while being directly accountable for what happens in the classroom.
That’s where this book is written from. Not because teachers know more about AI than the other three groups — they often don’t. But because the teacher’s position is the one where all of it intersects, with real daily stakes and no luxury of abstraction. For admin, AI is mostly a governance problem. For students, it’s mostly a resource. For coaches, it’s mostly a professional development question. Teachers are the ones who have to hold all of it at once — and still figure out what to do on Monday.
Most of the teachers I know fall into one of two camps. Some feel genuinely threatened — worried that AI makes their role obsolete, that students will use it to skip the work of learning, that they’ll be left behind professionally. The fear is real and it’s worth taking seriously, because underneath it is usually a genuine question: if AI can draft lesson plans, write feedback, and produce decent essays, what exactly is my job? That’s not a paranoid question. It’s the right question, asked badly. The answer isn’t that teaching becomes obsolete. It’s that the parts of teaching that can be automated probably should be, and that what’s left is more important and more human than the parts we’re handing over.
The other camp treats it as a novelty. They’ve been to the professional development session where someone demonstrates ChatGPT for twenty minutes, a few people laugh at the output, and everyone leaves mildly entertained but unconvinced. Or they tried it once, gave it a vague prompt, got something generic, and concluded: not ready yet. The problem with both experiences is that they’re testing the tool at its worst — with minimal context, minimal specificity, and minimal understanding of how to get something useful out of it. You wouldn’t hand a new teacher a class of 30 students with no preparation and conclude from the result that teaching is ineffective. The same logic applies here.
This book is written for teachers who are already using AI seriously, or who are curious enough to want to understand it before they decide. It is not trying to convince skeptics — but if you’re skeptical and still reading, that curiosity is probably enough. The ideas here don’t require you to have already committed. They require you to be willing to think carefully about what teaching is for, and whether AI changes the answer. If you’re in that place, you’re in the right place.
The shorthand I sometimes reach for — “AI is just a tool, like a calculator or Google” — is well-intentioned but it undersells what’s actually happening. The calculator comparison is especially worth examining, because people use it to imply that AI is neutral in the same way arithmetic is neutral. But with a calculator, the student still has to understand what operation to perform. They have to frame the problem. With AI, the tool makes framing decisions. It selects what to include, what to emphasize, what structure makes sense. That’s categorically different. It’s not retrieving an answer; it’s making a judgment about what the answer should look like. The “just a tool” framing papers over that distinction. And it’s a distinction worth taking seriously.
What I’d argue for instead is something harder to sustain but more honest: awe and humility. Not uncritical enthusiasm. Not alarm. Awe, because this technology is genuinely strange and genuinely capable, and the teachers who dismiss it as a passing novelty are making a bet their students are not making. Every student in your classroom has already decided that this matters. The question is whether you’re part of the conversation about how they use it, or whether that conversation happens without you.
Humility, because AI is also deeply limited in ways that are easy to miss when you’re only ever watching it succeed. It confabulates. It flattens. It produces text that sounds authoritative and is sometimes wrong. It has no idea who your students are, what they struggled with last Tuesday, what the room feels like when the energy shifts. I didn’t arrive at humility immediately — I had to get burned a few times, catch plausible-sounding errors I almost handed to students, realize that my confidence in the output was outrunning my verification of it. The teachers who will use AI well are the ones who understand its limitations as clearly as they understand its capabilities. That’s the disposition this book is written from, and it’s the one I’d invite you to bring to it.
This book makes a specific argument about that conversation. It is not enough to be part of it — teachers need to shift from reacting to AI to designing with it. Students who complete work with AI assistance are not failing; they are doing what efficient people do with efficient tools. The question is whether what we ask them to do next requires them to be present in a way that AI cannot substitute for. That is a design question, not a discipline question. Most school policy treats it as the second.
Teaching AI-integrated lessons in this context isn’t just a matter of picking up the right tool. It requires thinking differently about access, about culture, about what it means to learn, and about your own role in a classroom where students may reach for any advantage available — not out of laziness, but out of a very real, very human fear of falling behind.
This is a gap in the existing literature. Books and guides written for international teachers in China tend to address culture or curriculum, not technology. Books written about AI in education tend to address technology, not culture or curriculum — and almost never the specific realities of teaching behind the Great Firewall.
I should be honest about what this book won’t do. It won’t tell you that AI is the future of education — that prediction has been made about every technology since the overhead projector, and the results have always been more complicated than the promise. It won’t hand you a system that runs itself. And it won’t smooth over the genuine tensions in this work, because those tensions are real and pretending otherwise would be a disservice to you and to your students. And it won’t try to convince you that keeping AI out of your classroom is either possible or desirable. The question isn’t whether students use AI — it’s what you do with the fact that they do.
There is also a more fundamental reason to take this seriously: it is, simply, the job. Teachers have always been in the business of preparing students for a world the students can’t fully see yet. That has meant updating what preparation looks like with every generation — new economies, new civic demands, new technologies. AI is the update of this moment, and it is a significant one. The teachers who take it seriously, who actually learn how this technology works and what it does to learning rather than just reacting to headlines, are doing what good teachers have always done: getting ahead of the world their students are about to enter.
The complication — the one that makes this genuinely hard, not just unfamiliar — is that students are still becoming. They are in the middle of developing the capacities that would make them equipped to use AI well: the ability to evaluate a claim, to sustain effort on a hard problem, to think ethically about what a tool should and shouldn’t do for them. AI used without thought shortcuts exactly that developmental work. The teacher’s job in this moment isn’t to ban the tool or hand it over unchecked — it’s to understand it well enough to know when it serves student development and when it substitutes for it. That’s a harder job than anything a policy document covers. And it’s the one this book is trying to help with.
There’s something else worth sitting with. We talk a lot about students still becoming — about the developmental work that AI might shortcut or protect or accelerate depending on how it’s used. What we talk about less is that the AI itself is still becoming. The tools your students are using today are not finished products. They are intermediary stages in a technology that is developing faster than anyone can fully track — including the people building it. The AI that scared Anthropic’s own researchers into withholding a model from public release was not an aberration. It was the technology doing what it does: growing past what anyone had prepared for. When you introduce a student to an AI tool, you are not introducing them to a hammer. You are mediating a relationship between two developing minds — one human, one not — in a world that is still figuring out what that relationship should look like. That is a different job than tool instruction. And it’s closer to the real one.
I’ve taught in international schools long enough to have made most of the mistakes this book will warn you about, and to have had most of the breakthroughs it will point you toward. My goal isn’t to tell you that AI is the future of education. It’s to give you a framework for teaching when AI is in the room — one that starts from what you know about your students that the software never will, and works outward to the design decisions that actually make a difference.
How This Book Works
A note on scope. AI is moving fast enough that any book built around specific tools or capabilities will be outdated before you finish reading it. This one isn’t that. The tools and models I describe here will change — some already have. What won’t change is the framework: what does learning actually require, and how do you design for it when AI is in the room? That’s the question this book is organized around. It’s not a guide to what AI can do right now. It’s a way of thinking about what you, as a teacher, should decide to do with it.
There’s one more idea I want to name up front, because it runs through every chapter of this book. AI, in our classrooms, works like a diagnostic. It exposes what was already there — the pressure our students are under, the expectations we never fully explained, the parts of our instruction that weren’t quite holding, the structural gaps our schools have been tolerating for years. It doesn’t create most of the problems people blame it for. It makes them visible. Much of what follows is an attempt to take that visibility seriously.
Every chapter follows the same three-layer structure, because that’s how I think good professional reading should work.
It starts with a story — a real moment from a real classroom. Not a case study dressed up in anonymous language, but an actual scene, with actual stakes. This is where the ideas live before they become theory.
Then it moves into substance. Each chapter engages seriously with the research — what studies say about AI and learning, about Chinese education, about academic integrity, about feedback and assessment and motivation.
And then it closes with something you can use next week. A specific tool. A prompt. A classroom activity or redesigned assessment. That’s the part that connects the thinking to the doing.
The book moves through four sections.
Part One builds the foundation: what AI is doing when it generates text, what that means for learning, and what it means to teach in an international school — particularly in China, where the tool landscape looks different from anywhere else in the world.
Part Two is about the people in the room: the specific cultural pressures on students in international schools — especially in China — and the design frameworks that make it possible to build a classroom that actually works for the students you have rather than an imagined generic student.
Part Three is the field guide: lesson planning, feedback, assessment, differentiation, and teaching students to use AI well — each chapter structured around a real classroom problem with a practical set of tools at the end.
Part Four steps back to the institutional and ethical questions: school-wide alignment, academic integrity, and what teaching is for when the technology in the room is genuinely impressive.
You don’t have to read it front to back. If you’re mid-year and need something for Monday, go straight to Part Three. If you want the thinking before the doing, start with Part One. If you’re working on school-wide policy, Part Two and Part Four are where you want to be. If you just want to know how to use AI with students who are multilingual and operating under serious academic pressure, Chapters Three, Five, and Six are a short curriculum in themselves.
But wherever you start: take your time with the stories. That’s where the real stuff is.
A Note on Process
This book was written with AI as a collaborator. I used it the way I’d encourage any teacher reading this to use it — to generate structure when I was staring at a blank page, to help me synthesize research I’d read but hadn’t yet organized, to draft sections I then revised heavily, to check citations, and to think out loud with when I wasn’t sure what I was trying to say.
Every classroom story in this book is mine. Every pedagogical judgment is mine. Every revision that made something sound like me instead of like a language model is mine. The ideas came from teaching — from all of it, but especially from teaching here, where the questions this book asks have never felt theoretical.
I’m telling you this because the disclosure matters, and because the alternative — writing a book about honest AI use while hiding my own — would be exactly the kind of contradiction I’m asking you to help your students avoid.
Let’s begin.
