"The AI Doc: Or How I Became an Apocaloptimist" opens in theaters today. Director Daniel Roher — who won the Oscar for best documentary with Navalny in 2022 — spent months interviewing Sam Altman, Dario Amodei, Demis Hassabis, and dozens of other people at the center of the AI explosion.
His conclusion: the AI economy is "propped up by a Ponzi scheme."
That's a bold claim from someone who spoke to nearly every major player in the industry. I think the film raises questions worth taking seriously, even if you disagree with the framing.
What Roher Actually Saw
Roher and codirector Charlie Tyrell premiered the film at Sundance 2026, which had moved from Park City to Boulder. Roher didn't hold back in interviews — Vanity Fair published his unfiltered takes, and they're something.
On Sam Altman: Roher described him as "a machine" and compared him to the AI systems he builds. Whether that's a compliment or an insult depends on how you read it.
On Elon Musk: Musk agreed to participate in the documentary, then ghosted. Never showed up for the interview.
On Mark Zuckerberg: "He's like a lizard man." (Roher isn't subtle.)
The film's core argument isn't that AI doesn't work. It's that the economic structure around AI — the valuations, the infrastructure spending, the promises of returns — doesn't match what the technology actually delivers today. That gap, Roher argues, functions like a Ponzi scheme: early investors get paid by later investors, and the music eventually stops.
The Apocaloptimist Paradox
Roher calls himself an "apocaloptimist" — apocalypse plus optimist. He believes AI will change everything, possibly in dangerous ways, but he's not in the doomer camp. He thinks the truth about AI sits somewhere between "this will cure cancer" and "this will end civilization."
That's a more nuanced position than the title suggests. The film apparently treats AGI (artificial general intelligence) as a real possibility while questioning whether the current path to get there makes any economic sense.
For developers, this is the tension we live in daily. The technology is remarkable. GPT-4, Claude, Gemini — these models do things that seemed impossible three years ago. But the business models around them are still finding their footing. OpenAI reportedly lost billions in 2024. Most AI startups are wrapper companies with no moat. The hardware buildout dwarfs the revenue it generates.
Roher didn't invent these observations. He just put a camera on the people making the promises and asked uncomfortable questions.
Why Developers Should Care
This film matters because it shapes public perception. If mainstream audiences walk out of theaters thinking AI is a scam, that affects funding, regulation, hiring, and the tools available to us.
But it also matters because some of the critique is fair. The AI industry has overpromised repeatedly. Self-driving cars were supposed to be everywhere by 2020. AI agents were supposed to replace entire workforces by 2025. The gap between demos and production-ready systems remains wide.
Roher isn't wrong that there's a bubble component to AI investing. Whether it's a Ponzi scheme is a different claim — and a more inflammatory one. Ponzi schemes are fraud. AI spending might be irrational exuberance, which is a different animal.
Either way, the film opens today. It's got Oscar-level production, explosive interviews, and a thesis designed to provoke. Worth watching and forming your own opinion.
The AI industry could use a little more skepticism from people who've actually talked to the principals. Even if you think Roher's wrong, the fact that he got Altman, Amodei, and Hassabis on camera and still came away calling it smoke and mirrors — that's worth sitting with.
Top comments (0)