‘The AI Doc: Or How I Became an Apocaloptimist’ Team Explains Their 3-Year Journey Making Film, Interviewing More Than 40 Experts and More

Making a documentary about AI presented several filmmakers with a dilemma: How do you transform a subject better suited for a podcast or book into compelling cinema?

“This movie did not want to be a movie,” director Daniel Roher told an audience at the Sundance Film Festival’s “Anatomy of a Doc” panel on Jan. 23. “There was nothing here innately that was like, ‘Oh, that’s cinematic, that’s cool, that’s going to be great in a film.'”

Roher, who previously directed the Oscar-winning “Navalny,” compared the challenge to the scene in “Apollo 13” where engineers must fit “a square peg into a round hole.” What began as a planned one-year production stretched to nearly three years as the team faced a technology that was evolving daily.

“The AI Doc: Or How I Became an Apocaloptimist,” moderated by Variety senior entertainment and media writer Angelique Jackson, explores the intersection of AI development and humanity through an unexpectedly personal perspective. When Roher and co-director Charlie Tyrell learned they would both become fathers in early 2024, they anchored the film around that question: What kind of world will our children inherit?

“We were no longer people that aren’t destined to be parents,” Tyrell said. “If we were both still probably single guys without children, we would’ve had a really negative film that would just be called ‘Apocalypse.’ But we decided to call it this instead.”

The title itself sparked extensive debate among the filmmakers. Producer Diane Becker noted they cycled through more than 200 options — ChatGPT included — before landing on “apocaloptimist,” a made-up word that embodies the film’s philosophy of avoiding both doomsday scenarios and unchecked optimism.

“It’s not just a word, it’s a way of life,” Roher explained. “In a world that’s asking us to embrace visions of the apocalypse or lean into this unbridled optimism, we are saying no, there’s a third path.”

Producer Ted Tremper defined the term more practically: “Your emotions are the gas for the car, but you still need to steer. Being an apocaloptimist means understanding that there’s still a chance we can make it through all this if we change the default path we’re on.”

The filmmakers interviewed more than 40 experts on camera, generating 3,300 pages of transcripts. Tremper developed over 100 contacts inside and outside major AI labs, sometimes conducting pre-interviews lasting up to 20 hours.

“The secret sauce of the documentary is that we very unapologetically asked any question that any person who has never heard of this at all would ask,” Tremper said. His favorite question to every subject: “How do you think we’ll fuck up making a documentary about AI?”

Securing interviews with AI company CEOs required diligent preparation. “You literally listen to everything they’ve ever said. You read everything they’ve ever written publicly,” Tremper explained, comparing his research setup to the wall of note cards and red yarn in “A Beautiful Mind.”

The strategy involved interviewing people the CEOs had referenced in their work, building credibility that eventually impressed the PR teams at OpenAI, Anthropic and Google DeepMind.

To combat the challenge of making abstract concepts visual, the team leaned heavily on animation and stop-motion work. Tyrell’s crew recreated Roher’s Los Angeles art studio — an 8-by-20-foot shack in his backyard — in Toronto to serve as the animation backdrop.

“We want to be antithetical to the digital space of AI. We want to be handmade,” Tyrell said. The painstaking process yielded about 15 minutes of animation at a rate of four to seven seconds per day.

Editor Daysha Broadway spent nearly a year searching for the film’s structure, later joined by Davis Coombe, who had cut “The Social Dilemma.” The collaborative process involved producers, editors and directors all contributing across traditional roles and boundaries.

“It was really annoying having all those people,” Roher joked. “But literally in order to make this movie, that’s what we had to do. We had to reinvent a system.”

Tremper maintained levity during the grueling production by renaming their story outline document “Bucket Town, USA,” dubbing himself mayor and associate producer EJ Likes the sheriff. “This is what happens when you’re in mental prison and you need to find yourself a way to be able to go to work every day,” he said.

The film concludes with a pointed disclaimer: “This work may not be used to train AI.” Tremper emphasized the importance of such declarations even as the technology makes enforcement difficult.

“You do need to use your voice to push back, and you do need to celebrate the people who are trying to do it well, even if they’re failing,” he said, citing Adobe’s efforts to build models from licensed material as an example of companies attempting ethical AI development.

Tremper urged audiences to take action in their own spheres, recounting how he confronted his doctor’s office about using an undisclosed AI voice system for appointments. “The next time I called, that fucking robot was gone,” he said to applause.

When Jackson asked if the panelists had become “apocaloptimists” themselves, Roher, Tyrell and Becker said yes.

“The AI Doc: Or How I Became an Apocaloptimist” releases March 27.