Roger Deakins Calls AI in Film “Not Cheating”—Here’s Why That’s a Problem

The quote landed like a drone strike on a film forum: “I don’t think [using] AI is cheating. As long as you have something to say, I don’t care what you use.” Roger Deakins said it. Not at a tech conference. Not in a paid sponsorship for Adobe. At a Q&A—probably in some velvet-draped theater where the air still smells of photochemical nostalgia. The GOAT of modern cinematography just gave every AI bro with a prompt-engineering certificate permission to call themselves an artist.

Paul Schrader heard it and likely nodded while typing his next ChatGPT-assisted script. George Miller, still licking his wounds from the backlash over his “AI is here to stay” comment, probably forwarded the clip to his agent with a note: See? Guillermo del Toro, meanwhile, kept carving his woodblocks and muttering “fck AI”* into his coffee.

Here’s the thing about Deakins’ position: it’s technically correct, artistically catastrophic, and completely on-brand for a man who’s spent forty years making the impossible look effortless. Of course he doesn’t care about the tool. This is the guy who turned a digital camera into a paintbrush for Skyfall, who shot 1917 to look like a single, unblinking stare into the mouth of war. His “something to say” bar is so high it has altitude sickness.

But that’s not how the industry will hear it. They’ll hear: the tool doesn’t matter. They’ll hear: imperfection is negotiable. And in genre filmmaking—my wheelhouse, horror and sci-fi—that’s a death sentence.

Horror lives in the mistake. The practical effect that globbed wrong. The puppet string you almost see. The lens flare that wasn’t planned but feels like a ghost. The Thing (1982) is perfect because Rob Bottin’s latex skin tore in ways no algorithm would permit. Alien is perfect because the xenomorph suit was so fragile it could only be shot from certain angles, creating a silhouette that still haunts fifty years later. AI doesn’t do “fragile.” It does “iterate until seamless.”

Deakins knows this. He’s too smart not to. His work with the Coen Brothers—No Country for Old Men, Fargo—is built on stillness that trembles. You can’t prompt-engineer that kind of tension. You get it by putting a camera in a room with Javier Bardem and a cattle gun and praying.

But his quote reveals a deeper, more uncomfortable truth: the craft is already gone for most people. The average viewer can’t tell the difference between ray-traced lighting and golden hour. They don’t care if that sunset was real or generated from ten million Instagram sunsets fed into a neural network. They care if the story moves them. And Deakins is betting—maybe correctly—that a moving story, even AI-assisted, is better than a dead one shot on 70mm.

That’s the pragmatism talking. The man who shot Blade Runner 2049 knows that movies are a business of illusion. He’s just extending the logic: if CGI replaced miniatures, and digital replaced film, why not AI replacing… everything else? The answer is in the seams. CGI looked fake for decades—still does, often. Digital has a cleanness that cinematographers spend millions to dirty up again. AI will have its own tell, its own uncanny valley. We’re just not sick of it yet.

The real danger isn’t that AI will replace artists. It’s that it will replace the struggle. The eighteen-hour day trying to get a practical fire to flicker just right. The argument with the gaffer about whether a shadow feels “too angry.” The happy accident that becomes the whole point of the scene. AI smooths all that into a prompt: give me a moody church interior, Rembrandt lighting, slight lens distortion, golden hour, sense of dread.

You get an image. You don’t get a memory.

Deakins’ defense assumes intent is immutable. That “having something to say” survives the frictionless path to saying it. But intent erodes when it’s never tested. Schrader might generate a script, but Schrader also wrote Taxi Driver by hand in ten days, mainlining caffeine and self-loathing. The tool shaped the voice. The constraint created the art.

I don’t think Deakins is wrong. I think he’s fifteen years too early. When the novelty wears off and every streaming thriller looks like a Blade Runner knockoff generated from the same three prompts, we’ll remember why imperfection matters. We’ll spend millions adding digital grain back into the perfectly clean AI image. We’ll invent plugins that simulate mistakes.

Until then, his quote is a Rorschach test. If you already believed AI is just another brush, Deakins validated your LinkedIn post. If you believe art is the wound, not the bandage, you heard a master pragmatically surrender to a future he won’t have to shoot.

---

What Deakins’ Quote Actually Means

“Something to Say” Is a Moving Target
Deakins sets the bar at intention, but intention without constraint is just a mood board. The quote works for masters; for first-timers, it’s a license to skip the apprenticeship.

The Industry Hears What It Wants
His pragmatism will be weaponized. Producers already circulated the quote in Slack channels about budget cuts. “Deakins said it’s fine” is the new “Fincher storyboards everything.”

Genre Suffers First
Horror and sci-fi—the genres I live in—depend on tactile wrongness. AI generates perfect wrongness, which is just… nothing. You can’t be uncanny if you were never human to begin with.

Imperfection Isn’t a Bug
Deakins’ own best work has artifact—lens flare in Skyfall, motion blur in Sicario. AI treats imperfection as a filter you apply, not a mistake you earn. That’s the difference between a scar and a tattoo.

The Real Divide Is Age
Schrader (78) and Miller (79) embrace AI. Del Toro (59) rejects it. Deakins (75) takes the middle path. The pattern: filmmakers who’ve already proven themselves are willing to let the next generation automate the struggle they survived.

---

FAQ