Astana's AI film festival ditched the human versus AI bracket

I read about this at the Anchor, on paper, with Ray topping off a glass I hadn't asked him to top off. Eight thousand and sixty-seven films. A hundred and twenty-five countries. And not a single column anywhere asking what percentage of any of them a machine had made.

That's the whole story, if you want to leave now. Stick around and I'll tell you why it matters more than it looks like it does.

For two years this beat has run on one question: how much of this did the machine do. Every festival, every panel, every awards ceremony with a laptop and a projector has wanted a number. Fifty-one percent AI, sixty percent, entirely synthetic, human-assisted. It made sense when nobody trusted the tools. You count the dosage because you're afraid of the medicine. But a rule built to police a fear outlives the fear by about eighteen months, and then it's just paperwork.

The case for the bracket

Sparknify, the San Francisco outfit behind the Human vs. AI Film Festival, held its 2026 premiere on September 26 at the Delancey Street Screening Room and announced winners the next day. We covered that premiere when it opened. Sparknify's own release says the festival drew more than 3,000 submissions and put 32 shorts into competition, 16 made by humans and 16 made with generative tools, sorted into eight emotion categories: joy, sorrow, intensity, serenity, love, hate, desire, disgust. A grand jury of more than two hundred film-school professors and scholars gave the $3,000 Sparknify Humanity Award to a human-made film, "Down The Hatch," directed by Alexandra Lukina. A separate AI innovation jury gave Best AI-Generated Film to "BENCH," directed by Mooshin Rhee of South Korea's STUDIO aim.

There is a real argument in that structure and it deserves saying plainly: the two disciplines are not yet the same discipline, and a festival that judges them side by side, in the same category, with the same jury, might just be flattering AI work that is not ready to survive contact with a hand-drawn frame. Sparknify's own account of "Down The Hatch" leans on exactly that, the "expressive power of simple lines drawn by human hands," a thing that "can be fully encoded and decoded only through a shared understanding of what it means to be human." Fine. Brackets protect the thing that is newer and weaker. That is not cowardice. That is how you let a young fighter build a record before he gets thrown in with the champion.

But a bracket answers a question the audience stopped asking a while ago.

Why Astana skipped the bracket and won the argument

The Astana AI Film Festival, AAIFF 2026, did not ask filmmakers to declare a percentage, and it did not build a human column and an AI column. It built one competition and let 8,067 films from 125 countries fight it out inside it. According to Kazinform, reported by Anadolu Agency on September 17, the festival's co-founder and board chairman Almaz Zhali put the growth this way: "Just a few months ago, we were launching an international competition, and today we have received more than 8,000 entries from 125 countries." The United States led with 1,395 entries, Kazakhstan came second with 829, then India with 702, China with 550 and South Korea with 349. About 4,756 individual filmmakers made those films using more than 80 different generative models, and 3,138 of them tackled the festival's own theme, "The Future Worth Living In," while another 4,929 went into an open competition spanning 19 genres. We had already put the size of these prize pools next to every rival competition; what changed this week is that the entries are in and the festival opens on October 1.

None of those numbers describe a purity test. They describe a marketplace of finished work, judged as finished work, the way Sundance does not ask a director what lens he shot on before it decides whether the film is any good.

That is the tell. A festival that still separates its slate by origin is still treating the tool as the subject. A festival that stopped asking, and ran eight thousand entries through one funnel, has moved the subject back to where it belongs, on the screen. Sparknify's bracket is an honest, well-run answer to last year's question. Astana just is not asking it anymore.

Here is what a creator does with that. Stop putting your workflow in the pitch. Nobody handing out a real prize this fall cared how many models you touched. They cared whether a short about an old man secretly training for a school sports-day race made two hundred professors and a jury of strangers feel something. That is the whole bar now, and it was always going to end up being the whole bar. The festivals that keep counting your inputs are telling you, gently, that they do not yet trust your output. The ones that do not ask are the ones already judging you like a filmmaker.

I finished the beer. Ray did not ask what I thought of it. He never does. But eight thousand and sixty-seven strangers already told him.

More on AI films and the festival circuit

• San Francisco festival screens films without saying which are AI - the Sparknify premiere, September 26
• The 8 biggest AI film prize pools, ranked - what every major AI film competition actually pays
• Austin AI Film Festival returns to a city split over AI - the venue and the backlash
• Grok-and-Seedance short wins Reply AI Film Festival in Venice - the first edition's winners
• Six AI film contests with money attached close by the end of October - the deadlines still open

Read the Astana AI Film Festival entry figures as reported by Anadolu Agency, and Sparknify's own 2026 winners announcement here.