Prompt & Circumstance #1

Directing Is Prompting—It Always Has Been

I spent years as a film director before I ever typed my first AI prompt. Those years taught me that the heart of directing isn't yelling "Action!" or pointing dramatically at monitors—it's communication. You’re coaxing emotions, rhythms, and nuances out of human beings through carefully chosen words. You’re painting scenes through suggestions and corrections—an intangible art form that looks like nothing much from the outside but changes everything in the final result.

Filmmakers aren't taught to simply shout orders. Film school has entire classes dedicated solely to teaching directors how to communicate effectively with actors, crew, and creative collaborators. You quickly discover that vague feedback—“Make it better,” “Just do it differently,” or the dreaded “Can you be more intense?”—doesn't produce great results. Instead, you learn the subtle art of precision prompting: “Try it like you know the person is guilty,” “Say it like you know you are lying, but if you reveal yourself, you will be killed,” or Tarantino’s famously quirky direction, “Do it like you just remembered it.” These prompts might seem trivial, but they transform performances. They turn flat dialogue into iconic scenes.

When skeptics dismiss prompting AI as “just typing,” they’re misunderstanding the essence of creative leadership. The reason I love generative AI is so much of it is driven by natural language prompting. Allowing anyone who can speak or write natural language to become a Prompt Engineer. But this doesn't mean, everyone will be an expert Prompt Engineer. Prompting isn’t pushing a button and watching magic happen—it’s exactly the same act of guiding vision, tone, and performance. Just as a director doesn’t physically handle every camera or perform every role, a prompt engineer doesn’t manually render every pixel or type out every narrative beat. They shape the intention, the style, the feeling, and ultimately, the outcome. The tools differ, but the brain, instinct, and artistry remain the same.

Prompts Are the New Directing Notes

Consider directors like Greta Gerwig, Christopher Nolan, Jordan Peele, Ryan Coogler, or Quentin Tarantino. Each has a distinct style and signature, yet none of them personally operates the cameras, sews the costumes, or builds the sets. They provide detailed prompts—visual references, tonal suggestions, carefully phrased instructions—that inspire cinematographers, designers, actors, and editors to manifest their unique visions. In film circles, we revere directors precisely because of their skill in communication. They’re not celebrated for their ability to hold a boom mic but for the vision that animates an entire creative team.

Why then, when this same act of communication shifts to AI, does it suddenly become controversial or diminished? Prompting an AI effectively demands more than randomly entering words. It’s an art form—one that requires deep understanding of how subtle variations in language drastically influence outcomes. For example, AI platforms like ChatGPT explicitly document how prompt quality shapes results, noting small tweaks can profoundly alter style and nuance. Just as a director might adjust lighting slightly or ask an actor for a subtle change in intonation, prompting AI requires similarly refined judgment and iterative feedback. A tiny wording shift in an image-generation prompt—from “dramatic lighting” to “cinematic shadows”—can completely alter the mood, composition, and storytelling of an output.

Let me illustrate this with a real example from my own film, EAT. (Spoiler alert.) The final shot of the movie shows the main character, Novella, cutting her still-beating heart from her chest, taking a bite, chewing, swallowing—and then dying. It’s the crystalizing image of the entire film, the moment everything builds toward. We shot it a few times, and everyone on set thought we had it. The lighting was perfect. The framing was claustrophobic in just the right way. Meggie’s performance and makeup were flawless. The practical effects were stunning. Everyone was speechless.

Everyone except me.

It wasn’t right. Not the performance, not the image, but the tone. The emotional signature of the moment was completely off. This wasn’t how the movie was supposed to end—not in terms of plot, but in terms of feeling.

So I walked from the monitors into the cramped kitchen where we were filming and gave Meggie one note—one sentence. I told her: “Smile as you swallow.” You can actually watch this happen in real-time in our webseries. I have it queued up to the exact moment in the video below.

But that was it. A tiny, simple adjustment. And it changed everything. It reframed the entire scene and retroactively colored the whole movie. Suddenly it wasn’t just grotesque—it was transcendent, tragic, and weirdly joyful. That one prompt shifted the emotional core of the film.

That’s what prompting is, too. A single phrase, delivered with intention and vision, can transform the output entirely. Whether you're talking to an actor or an AI, the principle is the same: know what you want to feel, and say the exact right thing to make it real.

Working with these tools daily, I’ve learned that prompting is just like any other form of directing. Some prompts require a lot of work—like sitting down with an actor weeks before the shoot to unpack the scene and build emotional nuance. Others are simple, like asking an actor to walk across the room and open a door. One takes more time, the other less—but both are valid, and both are essential. Complexity doesn’t define artistic value. What matters is whether the direction, however big or small, serves the vision.

AI skeptics often dismiss prompt engineering because it seems easy at first glance—anyone can type, right? But consider this: How many people truly excel at directing films? The skill isn't holding a megaphone; it’s effectively translating ideas into execution. Good prompting is hard because it demands taste, instinct, and critical judgment. AI doesn’t magically compensate for poor prompting; if anything, it magnifies your shortcomings. A bad prompt creates a lifeless output, just like bad direction produces a lackluster film.

As Ben Affleck once said “AI is a craftsman at best … Craftsman is knowing how to work. Art is knowing when to stop.” He’s right—AI doesn’t know when to stop. It doesn’t know what’s too much or too little. It has no instinct for when something feels finished. But that’s not the AI’s job. That’s the prompt engineer’s. Just like a director calls “Cut! We're moving on!” when a take finally lands, the prompt engineer decides when the output hits the mark. AI might not have taste—but we do.

Prompting Is the New Auteurism

Ultimately, both directing actors and prompting AI rely heavily on trust—in your collaborators, in your instincts, and in your ability to refine and adjust until the result matches your creative intention. Prompt engineering is simply a new chapter in the old art of directing: guiding vision through language and suggestion, shaping the intangible into reality.

So the next time someone scoffs and says, “You didn’t make that—the AI did,” ask them this: “Do we say Christopher Nolan didn’t make Inception because he didn’t operate the camera? That Ryan Coogler didn’t make Sinners because he didn’t perform the music?”

Of course not. Because we understand that directing is authorship through collaboration. It’s not about doing everything yourself—it’s about knowing what everything should feel like and guiding a team, human or machine, to realize it. A director doesn’t need to act the scene—they need to shape it. A prompt engineer doesn’t need to draw the image—they need to summon it, to sculpt it from words with the same clarity and taste a director brings to set to build a scene.

Prompting is not button-mashing. It’s not a shortcut. It’s the art of specificity. The difference between “make it cinematic” and “frame it like a canted Dutch angle push-in at magic hour while the character silently processes betrayal”—that’s directing. And when you learn how to work with AI, you realize every word matters. Change one adjective, and your output veers into melodrama or cliché. Misjudge your tone, and the image or story falls flat. Sound familiar?

It should. Because that’s what directing has always been: making a million small, invisible decisions that add up to something singular. Something with voice. Something with soul.

If we’ve learned anything from a century of cinema, it’s that tools don’t make the artist—vision does.

The machine doesn’t make the art. The vision does.