Last year, filmmaker Paul Schrader—the director of Blue Collar, American Gigolo, and First Reformed, and writer of Martin Scorsese’s Taxi Driver—issued what seemed like the last word on artificial intelligence in Hollywood filmmaking. A few days after the release of Denis Villeneuve’s sci-fi blockbuster Dune: Part Two, Schrader asked his Facebook followers: “Will Dune 3 be made by AI? And, if it is, how will we know?”
Schrader is well regarded not only as a director, but one of cinema’s top-shelf curmudgeons, quick with a wry burn or baiting shit-post. But his Dune tweet seemed like more than another provocation. It spoke to a mounting feeling among many filmgoers, myself included: that Hollywood had stooped to producing sleek, antiseptic images so devoid of personality that they might as well have been made not by a living, breathing, thinking, feeling artist, but by a computer.
Most generative AIs “train” on existing troves of man-made images. With Dune, the opposite seemed true. It appeared as if Villeneuve was training on AI conjurations, screensavers, and glossy desktop wallpapers. (In fact, the film used “machine learning” models to relatively modest ends.) Still, it got me thinking: Is there an actual AI aesthetic? Do video generators powered by AI share a set of artistic ideas, or values, common among their output? Or, even more basically, can AI video generators have ideas, or values, at all?
My initial hunches here were … a) no; b) no; and c); no, of course an AI could not have “ideas” or “values,” which are the exclusive province of human artists, and human beings more generally. A toaster does not get a notion to warm up your bread or bagel, and then follow through with it. Nor does it care about how it does so. It merely executes a set of routinized, mechanized functions related to the warming (and eventual jettisoning) of breads, bagels, and other toastables. Why should generative AI be any different?
To test these premises (and my own rather dismissive conclusions) I trekked to a theater in New York to take in a program of 10 short films from the 2025 AI Film Festival.
The AI Film Festival is backed by Runway, a New York–based AI company offering “tools for human imagination.” Among those tools are image and video generators allowing users to create characters, sets, lighting schemes, and whole immersive scenes. With its Gen-4 software, users can theoretically create a whole movie—or something vaguely approximating one, anyway.
“We were all frustrated filmmakers,” says Runway’s cofounder, Alejandro Matamala Ortiz, of he and his partners, who met as grad students enrolled in the Interactive Telecommunications Program (ITP) at NYU’s Tisch School for the Arts. “We wanted to build the tools that we wanted to use.”
The film festival was born of a further desire to help legitimize those same AI tools. A gala screening held earlier this summer at New York’s prestigious Alice Tully Hall at the Lincoln Center (home to the New York Film Festival and year-round programming) saw filmmakers and technologists gather to watch the crème de la crème of a technology typically written off for producing mere “slop.” The festival format, Ortiz says, serves to “bring people together.” Now, that same gala program is touring Imax cinemas around the country, for a limited engagement.
As with any billing of 10 shorts made by 10 different filmmakers, the quality is a bit scattershot. The program begins promisingly enough with Maddie Hong’s Emergence, an immersive nature documentary “shot” (and narrated) from the POV of a butterfly larvae hatching from a chrysalis. With its bold pastel color palette, the rotoscope animated More Tears Than Harm, by Herinarivo Rakotomanana, superficially evoked the work of American primitivist painter Horace Pippin (who is one of my favorite artists). Simon Reith’s 6000 Lies is a rapid collage of gestating human fetuses, followed by a photo of a fetus burial site. In an abridged form, it might make an effective advertisement for a pro-life group.
Indeed, if there was anything like an aesthetic sensibility shared by the films it was a sense of commercialized gloss: rapid-fire edits, satiny, photorealistic images. A few, like Riccardo Fusetti’s Editorial and Vallée Duhamel’s Fragments of Nowhere, played like perfume ads for a fragrance an android might wear. The lousiest of the bunch was an anime short called RŌHKI - A Million Trillion Pathways, credited to a filmmaker named Hachi and IO. Beyond being wholly derivative, it highlighted the rather obvious shortcomings of the technology, like characters’ earlobes and shirt collars seemingly mutating in shape between scenes.
One filmmaker in the audience, Robert Pietri, came away mostly impressed by what he saw. “A couple of them were really pushing, and going where I think you should be going with this,” he says, “which is creating a cinema that you can’t create otherwise. I was excited by it.” He sees the weaker films as not being limited as much by the emerging AI toolkit but by the “limitations of the creators.” An AI it seems, cannot render away bad ideas inputted by the human beings plugging in the prompts. Well, not yet, anyway.
As something of a generative-AI skeptic, watching the program raised all kinds of questions. Some of these were pretty pedantic and boring. Like: Does standard movie theater etiquette (re: looking at one’s cell phone) apply during an AI film fest? You could imagine a computer filmmaker might actually like to see another little computer, lighting up in the dark theater, as if approvingly.
Other questions were a bit more existential—or, perhaps, ontological—relating to the very nature of so-called “AI art.” Even when these films were entertaining or nice to look at, I couldn’t help but feel a little tricked. Aren’t those qualities mere impersonations of real films, painstakingly made by real people? And so, aren’t even the “good films” still fundamentally bad?
There are other, less chin-stroking, considerations. Generative AI has drawn criticism for its massive draw on natural resources, with OpenAI CEO Sam Altman admitting that the proliferation of data centers is pretty much unsustainable without considerable developments in nuclear fusion. And AI’s encroachment on certain creative fields has led to concerns about mass layoffs in the film and video game industries. Runway’s Ortiz says he thinks of AI as similar to previous technological innovations rollicking creative fields. “Technology has brought some disruption in the job market,” he says. “But it also opens up new, new stuff.”
The technologists and “creatives” using and promoting these tools tend to frame the arrival of all this “new, new stuff” as something of an inevitability. Phoenix-based filmmaker Jacob Adler, whose Total Pixel Space took top honors at the AI Film Festival, seems to follow a similar philosophical track. “I am fascinated by the long arc of technological evolution that requires the flow of time, beginning with biological technologies such as self-replicating molecules, cell membranes, photosynthesis, nervous systems, eyes, brains, etc.,” Adler wrote in an email to WIRED. “AI is not a departure from nature, but a continuation of the fundamental evolutionary trend of biology learning to build more complex information-processing systems, now outside its own flesh.”
But flesh still has its defenders.
The AI battle lines increasingly seem drawn in terms that are unreconcilable. So it comes as little surprise that skeptics, critics, and champions of the old ways have been decrying the AI Film Festival from day one.
When Imax announced its partnership with Runway AI, responses among cinephiles were spiky. One X user replied,“Not watching anything made by clankers,” referencing the slur used to disparage robots in the Star Wars films. Actor Jared Gillman reposted Imax’s announcement on X with an image of Ethan Hawke in a suicide vest from First Reformed with the caption, “One ticket for the ai imax film festival please” (doubly appropriate, perhaps, considering Schrader’s stated views on the technology). The commingling of Imax and AI seemed a particularly egregious offence. Imax, after all, is a corporation ostensibly dedicated to showcasing the theatrical cinematic experience at the grandest and most imposing scale. And AI is regarded as, well, something else entirely. As one especially withering Redditor put it, “Imax and Runway AI Sign a Film Festival Deal to Show Dogshit.”
AI’s defenders (many of whom have a vested financial interest in the technology’s success) love claiming that many great leaps forward in human artistry have been met with similar resistance. And they have a point—to a point. Digital filmmaking challenged analog, celluloid filmmaking. The introduction of sound and color technology in cinema was regarded, at first, as a mere gimmick. Even very early critics and academics fretted that photographic media like cinema could never be art, precisely because they merely represented reality, instead of interpreting it.
With time, and plenty of counterargument, pretty much all these takes have been proven wrong. Adler’s own prize-winning short (which earned him a $15,000 cash prize and 1 million Runway AI “credits”) is a thesis film on this very idea. Total Pixel Space explores the notion of a hypothetical universe of colored pixels, exploring what its narrator calls “a process of discovery in which all of reality is already mapped.”
OK. But is all of reality “already mapped”? Is this technology merely being “discovered,” as one might discover, say, a river or a cool restaurant? Or is it being invented, and managed, by actual people making actual decisions, which have actual consequences? Perhaps AI is just the latest disruptive innovation riling up the haters and fuddy-duddies. Or perhaps machine-generated art constitutes not just another step in cinematic-technologic evolution, but a fundamental break from the basic, taken-for-granted idea of what it is to make art: that it involves skill, tremendous patience, considerable talent and, at very least, a human being positioned as its prime mover. AI is a difference (or “disruption”) not of degree, but kind—not the next step in a process, but a totally different thing. Apples and computer-generated oranges.
For Troy Petermann, a 15-year-old attending the New York screening of the Imax AI Film Festival with his family, AI is not a tool. It’s a threat. “AI is definitely an innovation,” says the aspiring filmmaker. “But innovation is the drug of humanity. We never know how to stop when it goes too far.”
Petermann’s reflections are refreshing, in large part because generative AI technologies are so typically pushed to people in his exact demographic: wannabe filmmakers with big ideas but little in the way of tools, money, or institutional support. He admits that AI technology may have tremendous upside in terms of its “analytical” capacities, like processing and synthesizing information. “When it comes to creative aspects,” he says, “we should just draw the line.”
Increasingly, those lines are getting blurrier. The AI Film Festival earned such pointed cinephile scorn in part because it was easy to single out as a conscious enemy of the seventh art. Runway’s Ortiz speculates that, for the festival’s next installment, the company may change the branding altogether. “I don’t think it will remain the ‘AI Film Festival,’” he says. “We do think AI is just going to be part of any process. Similar to other companies. Everyone is an AI company, and will be using AI in some way. AI will become just another tool within filmmaking.”
