Last week saw the official arrival of a whole new genre of filmmaking.

Or, at least, that’s how Sora is touting the rollout of its artist program. Sora Selects has arrayed 10 artists and artist teams who have used the artificial intelligence video generator to create short films that showcase the tool’s creative possibilities. OpenAI, the company behind the model, has invested $3 million into the initiative, with its team spending the past 10 months fostering a relationship with these artists.

“It’s a lot about understanding that this is an archive of human ingenuity and creativity, and it also requires practice and learning a new vernacular,” Souki Mansoor, the lead of the Global Sora Artist Program, told me about the initiative. “We wanted to really empower the artists who have become the most fluent in this art form to be able to share their knowledge and educate others.”

Boey, Dimensions (still). Photo courtesy of the artist and Sora.

Launched to the wider public early this year, Sora enables users to generate up to one-minute videos by entering a text prompt. A host of features allows them to iteratively edit or remix the videos, or to storyboard the scenes down to the last detail. Early access to the model, which opened in December, saw 1,000 sign-ups per minute, Sora’s product lead Rohan Sahai told me; at peak times, Sora handles some 10 generations every second.

“There’s just such a wide array of what people are doing with this, which is really awesome,” he said.

Steven Schardt, The Great Flood (still). Photo courtesy of the artist and Sora.

Among those people are creatives, a demographic that Sora has been built for and targeted at. Mid-2024, a group of artists was given free access to the tool, which, while prompting a scathing response with accusations of “art washing,” has also helped the company build a creative community that’s been key to its artist program.

“These are the early days for generative video. And one thing that we saw was a lot of people using the community as a way to learn and experience the tool,” said Joey Flynn, designer for Sora, about the platform’s user experience. “We thought that was really important to foster this idea of ‘we’re all in this new experimental art form together.'”

Adeline Mai, Ocean of Tears (still). Photo courtesy of the artist and Sora.

It’s from this participation that Sora Selects has drawn its first class, featuring artists Adeline Mai, BOEY, Boldtron, Boris Eldagsen, David Sheldrick, Isabelita Virtual, Steven Schardt, Panaviscope, shy kids, and the duo of Manuel Sainsily and Will Selviz.

These creatives had quite the outing at the Metrograph theater in New York, where their films were shown on January 30. Projecting these works on the big screen, Mansoor said, was a “magical” exercise, as was, presumably, the demonstration of Sora’s high-resolution output.

Sora Selects artists at the New York screening at Metrograph. Photo courtesy of Sora.

So, what have these artists made with Sora?

If you’ve ever laid eyes on an A.I.-generated video, you might know what to expect. Some of these films are riven with the surreal, disorienting morphing so telling of the generative medium. A living room transforms into a forest! You’re now in a nightclub—or are you? There’s a car… oh wait, no, it’s a bunch of clouds! These are good-looking videos that are thankfully brief enough to sustain that kind of dream logic.

David Sheldrick, 2AM (still). Photo courtesy of the artist and Sora.

More engaging are the films that offer a narrative throughline: Panaviscope’s Maybe I Got Carried Away, which tells the story of a woman whose penchant for balloons goes awry, and Schardt’s The Great Flood, which recounts the fictional flooding of Louisville, Kentucky, using 1950s-style footage and voiceover. They’re only let down by the model’s poor generation of crowd scenes and characters that don’t look like they’ve emerged from a stock photo library (providing some idea of the “publicly accessible material” that Sora was trained on).

Two compelling stories stand out. Isabelita Virtual’s Soledad, about a lonely specter who meets its mate, nails a haunting vibe that vastly surpasses its A.I.-generated medium. Meanwhile, the charming Florp’s Solar Vacation by shy kids, featuring a young alien’s school report of its trip to Earth, manages to make comic yet poignant hay out of the model’s output.

shy kids, Florp’s Solar Vacation (still). Photo courtesy of the artist and Sora.

“There were these beautiful images that were being made [by the model], and I think that you can excuse them a little bit more when they’re coming from an alien, especially a child alien,” said a member of the shy kids at the screening. “I thought it was appropriate for how the footage was making us feel.”

Then again, emphasis on “narrative” in this brave new world of generative filmmaking might seem limiting or even passé. As Sahai told me: “The temptation is to plant this technology on top of the existing industry and workflow as we know it, but I think that’ll feel silly in five years. It’s like, ‘oh yeah, of course this enabled this completely new thing and this new medium.'”

Manuel Sainsily and Will Selviz, Protropica (still). Photo courtesy of the artist and Sora.

Sainsily and Selviz, for one, are thinking beyond the big screen. They’ve broadened Protopica, their film that meditates on heritage and identity, into an interactive installation, wherein participants, wearing EEG headsets, will get to view variations on the film based on their brainwaves. “It’s the non-linear storytelling you see in the theme of the short film that we are now expanding towards,” Sainsily told me.

Of course, while A.I. can be generative, it is also extractive. Models have been trained on generations’ upon generations’ worth of human labor and creativity (OpenAI is currently facing several copyright infringement actions), if just so we can now create a short film in a week with a tool that could further impact artists’ livelihoods. However much Sora helps creatives realize projects without significant outlay, it comes at a cost.

Boldtron, Memory Fiction (still). Photo courtesy of the artist and Sora.

On Sora’s part, Mansoor acknowledged that the tool will inevitably disrupt the creative industry, but sees the “net benefits” in how it might usher in “a new economy, a new way of working.”

“I don’t know if we fully understand what Sora is capable of,” she said. “As our artists start to discover it with us, as we venture into the unknown, I think we’re going to get even more understanding of what mediums it can create and how it can influence people’s livelihoods.”

One artist who has tussled with the ethics surrounding the technology is Berlin-based Eldagsen, who first came into prominent view in 2023 when he submitted an A.I.-generated image to a photo competition and won. His Sora Selects entry, Measurement Is King!, is a jazzy, black-and-white vignette of besuited people sizing one another up with measuring tape, an effort to poke fun at the corporate practice of benchmarking. It’s also a side swipe at the competitiveness among A.I. companies.

Boris Eldagsen, Measurement Is King! (still). Photo courtesy of the artist and Sora.

His thoughts on A.I. revolve around its gray areas—from its copyrightability (a problem that is “outsourced to the users,” he said) to the lack of regulations in the space (particularly as the new U.S. administration launches a $500 million A.I. push). Using Spawning, he’s learned that his own work has been used as training data, a fact, he told me, that is simply “the reality.”

There are bigger questions too: “What are the implications of a future where most of the internet is generated? [There’s] the loss of authenticity, the potential for disinformation, as A.I. becomes the new super fuel for the social media engine.”

“We need definitions to be able to find the strength of A.I. and humans. What can A.I. do that’s good or better and faster? What can we outsource?” he added. “What is remaining for us?”