v1 - April 2026. This preface is a working draft, shared as part of building in the open. A revised version lands later this summer, and the final text publishes with the book in 2027.

"It's never been easier to create footage. But it's never been harder to make a film." Derrick Schultz said this at his artist lecture as part of the AI Greenhouse program last fall and it's been sitting with me since. The barrier to producing (moving) images, text, and sounds has collapsed. The work of making something that means anything, that holds together, that you're willing to put your name to, that knows what it wants to say and why: that has gotten stranger and harder, not easier.

This book comes out of a year of conversations. With students asking what they're allowed to use and what counts as their own work. With teachers trying to keep up with tools that change every few months, wondering what's worth teaching at all. With artists pulling in different directions: some embracing AI with open arms, others refusing on principle, most somewhere in the middle, working things out as they go. None of these people are wrong, exactly. They're all responding to something real.

Every tool we use changes how we think. A pencil is not a typewriter is not a camera is not a screen. So when we ask what AI is doing to filmmaking, or to art education, or to how young people learn to see the world, the honest answer is that we don't fully know yet, and we won't for a while. But it's doing something. It's already doing something.

This book is for people trying to teach and learn creative practice in that gap and holding a space for noticing what it's doing, together.

The shape of what's happened is bigger than "AI arrived in filmmaking." Production has collapsed in cost and effort. An image can be spawned in seconds from a few words. A voice, a style, a world. Tools that did not exist three years ago now exist and change every few months. Tools that do exist are opaque even to the people who build them. The legal, ecological, and labor conditions underneath those tools range from unclear to actively bad. And the field keeps calling all of it "AI" as if that were one thing, when really it's a bundle of both settled and emerging technological capabilities shifting how we make, how we see, how we pay attention, and how we teach.

Don't believe the hype. In October 2025, OpenAI launched its first big ChatGPT brand campaign, a series of cinematic 30-second films carefully crafted by a large team of creatives and shot on 35mm film, just hours after launching Sora 2 (their video AI). The companies selling AI as the inevitable future of creative work benefit from us believing that story but their own ChatGPT campaign suggests they don't fully believe it themselves.

And what worries us about all of this is rarely the technology in isolation. It's the values that produced it, the agenda it continues to push, the power structures that benefit from its rapid spread, and how all of those shape what the technology becomes.

What we've noticed, working with students and teachers this past year, is that each of these shifts is its own live question. Can you still call yourself the author of an image you prompted into existence? What counts as your work when a model trained on millions of other people's work does the heavy lifting? What are you doing to the planet when you generate a minute of video, and does it matter, compared to what? Can you trust what you see on a screen, and should you? How do you teach craft to people who may be persuaded to bypass some of the learning curve, and what parts of the creative process are worth outsourcing or struggling for?

None of these questions resolve cleanly. They resolve differently depending on who's asking, what they're making, and what they care about. Our approach has been to stop pretending otherwise.

Here's what we've come to believe. A lot of the confusion right now comes from treating AI as a tool to be mastered. It isn't really that, or not only that. A camera is a tool. A pencil is a tool. The thing that makes a camera useful is that you know roughly what it will do when you press the shutter. Generative models are closer to a slot machine than a camera. You input something, something comes back, and the relationship between input and output is statistical rather than mechanical. That changes what it means to work with one. It doesn't mean you can't get good at it. People are learning the material and making incredible things, both by taking tight control and by inviting randomness or unlikely behaviour. But it's a different kind of getting good, and pretending it's the same thing as learning a camera misses the point.

The instinct to hand students a list of tools they should know and have them check them off is the wrong shape of response. The tools will have changed by the time they graduate. What won't have changed is the underlying question: how do I decide whether to use this, why, and how do I defend that decision to myself and to other people?

AI is also a tool we think with, not just one we make with. Working with it changes how we form ideas, what we offload, what we keep, how we sit with not-knowing. There's an ongoing conversation in our circles about how people are or aren't using AI, who's "letting it think for them" and who isn't, framed as a binary. Stan Liguzinski, research coordinator at NFA, shifted the question for us: the more useful one isn't whether to let AI do our thinking, but how working with it changes our thinking. The term for what's at stake is something like cognitive sovereignty: the capacity to keep your own reasoning, intuition, and creative judgment intact while working alongside tools that will happily do those things for you. Recent studies from MIT and others on cognitive offloading suggest this isn't paranoid.

There's a related concern about how skills get built. A lot of senior practitioners developed their judgment, their taste, their instinct for when something is wrong, by working through tedious phases that AI now compresses or removes. There's a real risk that a generation growing up with these tools will be more productive output-wise but with thinner foundations, less practiced at the slow, friction-laden work that builds those instincts. We want to be careful with this argument. Every generation has worried about the next losing skills (handwriting with typewriters, mental math with calculators, navigation with GPS), and a lot of that worry was overblown. Some wasn't. We don't yet know which this is. We also want to flag that the friction we're talking about isn't always good: plenty of students who reach for AI early are doing so because traditional education was already failing them. The honest position is that this is a live question, and the students whose senior practitioners we'll need in ten years are working out their relationship to these tools right now.

So the book you're holding is built around that question.

Each recipe in here is a hands-on exercise an educator or artist has developed in their own practice and is sharing with you. Some are fifteen minutes, some are a full day. Some use AI tools, some deliberately don't. Some are about technique. Most are about something larger: how a machine sees an image, what it means to caption a photograph, what a film's rhythm looks like when you lay it out in space instead of time, what you can see about a film when the machine looks at it with you, what your own values are and where they bump up against what the tools want from you, or you from them.

What the recipes share is a pedagogical stance we want to be explicit about.

The first piece of it is that teaching in this moment has to be honest. Teachers are some of the most overworked and underpaid people doing some of the most important work in any society. None of them signed up to be AI experts and it would be unreasonable to ask them to become ones. The field moves too fast. So the teacher's role here is less about knowing all the tools and more about holding the space for learning, asking good questions, and drawing out what the room already knows. Film students in 2026 are navigating a landscape nobody fully understands yet, including their teachers, including the tool-makers, including the policymakers. Pretending otherwise is worse than admitting it.

The second piece is that critical AI literacy, in our view, isn't a syllabus to be covered. It's a practice to be sustained. A collective capacity to sense-make together as the tools, discourse, and landscape keep shifting. That phrasing is borrowed from AIxDESIGN, where it landed after a year of trying to say something more precise than the field's defaults. The mainstream AI literacy frameworks want to make this look like a skills inventory: know these concepts, use these tools, understand these ethics. That's useful as far as it goes, but it misses what the work actually feels like in a room full of twenty-year-olds trying to figure out whether they should use a tool that might put them out of work in three years and might be killing the planet while they do it. What's actually at stake is practice, value, and position. And the room is often smarter about it than you expect, if you offer a fruitful container.

The third piece is that this is social work. Critical thinking about AI doesn't emerge from reading the right articles, though that helps. It emerges from comparison, disagreement, and surprise. From seeing what a machine does with your image next to what your classmate does with the same image. From watching an AI film together and realizing you hate different parts of it for different reasons. Every recipe is built for a room of people learning together rather than a lecturer delivering information. Every recipe has at least one designed social moment, a specific structured interaction where the learning lands because it happens between people.