Will AI cinema ever stick?

Paul Schrader recently speculated that a savvy student will soon be able to create a 90-minute narrative in a few weeks, on no budget, without leaving home, without permission. Schrader is not fantasizing about gimmicks; he is pointing to a structural shift. If the tools for writing, shooting, editing, scoring, and even casting become prompt-driven, then the bottleneck of production dissolves. Why endure the ordeal of fundraising and coordination when a keyboard will do?

It is a provocative claim, yet not entirely unprecedented. I like the parallel with literature: anyone can write a book. Anyone can upload it. Yet there is still Penguin Random House ruling the world. There is still the The New York Times Best Sellers list.

The fact that production is democratized does not eliminate hierarchy. In fact, it makes it clearer how this value chain is organized, orchestrated, and what combination of factors need to be in line in order to deliver the goods to the viewer.

I see three questions, none being whether AI can generate feature-length films. What new agents will decide which of those films matter?

From Production Scarcity to Attention Scarcity

For most of cinema’s history, the central constraint was material. Cameras were expensive, crews were necessary, distribution was gated. Studios mediated risk. The so-called “star system” emerged partly as an economic solution: recognizable faces reduced uncertainty. When Meryl Streep appears on screen, she carries decades of accumulated trust. Her presence compresses doubt.

If AI collapses production costs, scarcity shifts from production to attention. We have already seen this with literature. The self-publishing revolution did not eliminate publishers; it intensified the value of curation. Penguin does not merely print; it filters, signals, packages, amplifies. The New York Times list is less a record of quality than an infrastructure of recognition.

The same logic will apply to audiovisual narratives. When everyone can generate a film, visibility becomes the premium commodity. Platforms, festivals, algorithmic feeds, and influencer-critics will function as the new studios. The power center moves upstream, toward selection and framing.

In that sense, the more radical change may not be the capacity to generate images, but the emergence of new curatorial authorities. YouTube may suffice, or it may fragment further into niche platforms designed specifically for AI-native cinema. The economics of permission will fade; the economics of prominence will harden.

Simulation and Suspension of Disbelief

Yet there is a deeper question beneath distribution: will we believe what we see?

Cinema has always required a contract of imagination. We agree to accept constructed worlds as emotionally real. But that agreement rests partly on a specific ontology: we are watching human bodies performing before a camera. Their tears are performed, but they are also embodied. A human face, in close-up, carries micro-movements accumulated through lived experience.

If a character generated by AI cries, is that sufficient for us to cry with them? Or is there an irreducible layer of recognition when we know that an actor has lent their body, their breath, their vulnerability to the role?

Consider ballet. If humans could naturally fly, the spectacle of elevation would lose its meaning. The art form is powerful precisely because we know the body is resisting gravity. The tension between capacity and limit creates beauty. In cinema, something similar happens. When an actor trembles, we sense the effort beneath the expression.

An AI-generated face may be flawless, but its perfection risks flattening the tension between body and constraint. The uncanny valley is not merely technical; it is existential. We are attuned to the knowledge that someone is risking something before us. Even in fiction, we perceive effort.

This does not mean AI narratives will fail emotionally. Animation has long moved audiences profoundly. We weep for drawings. But the emotional grammar of animation is stylized; it signals its artifice. Hyperreal AI simulations occupy a more ambiguous zone. They imitate the photographic trace while lacking the photographed body.

I am confident many animation fans will argue that an actor is entirely unnecessary. But considering the niche in which these animations historically play, and the fact that they never became the mainstream, and we may understand that for at least 100 years, the mainstream of the moving image does connect better with actors. Would it be different now? The question is not whether we can be fooled, but whether we care to be.

The Problem of Abundance

Schrader suggests that originality will determine value. In a world of infinite production, distinctiveness becomes the differentiator. This is persuasive, but incomplete.

Originality has always been necessary but rarely sufficient. What transforms a narrative into a cultural event is a coalition of forces: timing, endorsement, distribution, reputation, repetition. A novel does not become a bestseller solely because it is inventive. It becomes one because a network of institutions amplifies it, and a signal that reaches the mainstream as vanguard systematically gets worn out by the usual movement of market demand, which in turn demands more novelty. Whatever tools the creator decides to use, runs on moving ground, and needs to keep ahead of the curve.

In film, this network includes directors whose names carry symbolic capital. Darren Aronofsky experimenting with AI is not equivalent to a student doing so, even if the student’s work is equally compelling. Authorship functions as a brand. The democratization of tools does not erase symbolic capital; it often intensifies its importance. When technical barriers fall, reputation becomes the currency.

The Star System After the Star

We cannot ignore the star system. Audiences often seek not only stories, but encounters with familiar presences. Seeing an unknown but talented actress in a drama can be exhilarating. Yet seeing a figure whose face has accompanied us for decades carries another pleasure: continuity. The star is a stable reference point in a volatile world.

AI complicates this. Synthetic actors can be infinitely malleable, eternally young, endlessly available. But they do not age with us. They do not accumulate scandal, triumph, biography. The fascination with a performer lies partly in the knowledge that they have lived, erred, transformed.

A digital actor can simulate aging; it cannot have memories.

Perhaps the future star will be hybrid: a human performer licensing their likeness to AI systems, curating digital extensions of themselves. Or perhaps entirely new forms of stardom will emerge—creators whose primary medium is algorithmic orchestration rather than embodied performance.

Still, it is worth asking whether the presence of a living, vulnerable person is not a constitutive layer of cinematic meaning. When we watch a great actor, we do not merely watch a character; we watch a person navigating the limits of their body and psyche.

Platforms as the New Studios

If production is frictionless, platforms become decisive. The studio era controlled production and distribution. Streaming platforms reconfigured distribution. AI-native cinema may invert the equation: production everywhere, distribution as choke point.

The platform that can curate, recommend, and monetize AI-generated narratives at scale will shape taste. It may resemble YouTube; it may resemble a festival circuit; it may become a subscription-based “AI cinema” hub. Whatever its form, it will function as a legitimizing apparatus.

The analogy to publishing returns. Self-publishing did not destroy publishers; it multiplied manuscripts. The value shifted toward selection, branding, and trust. Readers still seek signals that help them navigate abundance. Viewers will do the same.

What Remains Human?

The deeper anxiety beneath these debates concerns displacement. If a student can create a feature film alone in a room, what becomes of the collaborative art of cinema?

Yet collaboration may not disappear; it may relocate. Instead of coordinating lighting crews and camera operators, creators will coordinate prompts, datasets, and model behaviors. Instead of directing actors, they will shape performance parameters. The craft will mutate, not vanish.

Still, there is a residue that resists automation. When we watch a performer cry, we are not only witnessing pixels arranged to resemble tears. We are witnessing a body under tension. Even if the tears are glycerin, the breath is real. The heart is beating.

If AI characters cry, we may feel something. But whether we feel the same intensity may depend on whether we sense risk—whether we perceive that someone, somewhere, has wagered themselves in the act.

In literature, the author’s body is absent from the page, yet we still value books intensely. The difference is that literature has never relied on photographic trace. Cinema, by contrast, was born from it. Its ontology is tied to the idea that something happened before a lens.

AI severs that tie. The image no longer guarantees an event.

This does not doom the medium. It transforms its metaphysics. The new cinema may be less about recording and more about generating. Its authority will not derive from capture, but from coherence and affect.

The student in their bedroom may indeed create a 90-minute narrative without permission. The real contest will be for belief and attention, courting institutions that will reassemble around curation. So the tears on screen may be synthetic (we’ve seen this before). But will they be perennial like cinema has proven to be, or disappear in the rain?