At some point in this decade, interviews with a certain kind of subject matter expert all began to end in the exact same way: The interviewer asks the guest, who spent the past 10 minutes telling us how a fast-approaching pandemic, climate disaster, or other calamity will definitely wipe out 90 percent of humanity, how they feel about the future. The guest says, “Well, I’m an optimist…” then gives one or two vague, unconvincing supporting reasons, and we cut to commercial. This, in broad strokes, is the narrative arc of “The AI Doc: Or How I Became an Apocaloptimist,” a new documentary on the nuclear-level impact AI could have on our future.

Hot off his Oscar-winning doc about Russian dissident “Navalny,” director Daniel Roher and his co-director Charlie Tyrell deploy an intently folksy and personalized series of explainer videos and expertly-lit and -photographed interviews with what a talk-show producer would call “real gets”: OpenAI’s Sam Altman, Anthropic’s Dario Amodei, Google DeepMind’s Demis Hassabis, and seemingly every other leader, innovator, commentator, or activist in the AI world. (Missing but not missed: Musk and Zuck.) Roher also makes the rather bold decision to center himself and his family in this discussion about what several of his guests call the biggest inflection point in human history. So if you feel that Robert Oppenheimer’s struggles around his security clearance is the most interesting thing about the atom bomb, this film may be for you. 

You can’t entirely fault Roher for this, since life did serve him an empathy-winning conceit—an imminent father contemplates the future his son will inherit. But your enjoyment of “The AI Doc” will depend on how charmed you are by the anxious, earnest, not-terribly-probing interviewer at the center of this film, as both its protagonist and narrative filter. Indeed, the film’s first segment is dedicated to Roher’s Bildungsroman, narrated in storybook mode by his wife, the filmmaker Caroline Lindy, who also often speaks as a character in one of Roher’s animated line drawings, relaying how young Daniel was born in the early ’90s, grew up with a camcorder in his hand, and used digital technology to make art and movies. Trouble in paradise comes with the couple’s early encounters with ChatGPT, which they learn can write screenplays as well as human beings.

Both the narrative thus far and the film’s handmade aesthetics—with Roher’s drawings and watercolors and Oscar-winning producer Daniel (Everything Everywhere All at Once) Kwan’s (probably CGI-generated) stop-animation—feel as if they were created in opposition to Apple’s tellingly tone-deaf ad for the new iPad, which took a DOGE Bro-like glee in destroying art supplies, musical instruments, and other human-beloved analog creative interfaces to empower one overpriced digital device (a decent synecdoche for ChatGPT). But after this, when the expecting father finds creative purchase in AI doomsaying, he invites some of its key exponents into his studio, confronts the terror of scale and iteration, and spirals out into the all-too-Hollywood sentiment expressed by the film’s sub-Strangelove subtitle “Or How I Became An Apocaloptimist.” 

Anyone who follows this stuff knows that the most outspoken figures in Silicon Valley are often obscenely rich and observably mentally ill: proudly ignorant of history and literature, prone to statements like Marc Andreesen’s claim that “400 years ago, it never would have occurred to anybody to be introspective” (Plato, the Bhagavad Gita, St. Augustine, Descartes, et. al. to the contrary), as well as openly contemptuous of democracy and/or women while seeking more and more centralized power. These people are noticeably absent among the technologists, academics, and risk-management experts who appear in The AI Doc and lucidly describe today’s unregulated arms race toward generative AI: an opaque technology that’s vastly smarter than all of humanity combined, constantly seeks power, operates at a scale inconceivable to human beings, and, per the beetle-browed computer scientist Eliezer Yudkowsky, poses a significant risk of causing “the abrupt extermination of the human race.” It’s very hard to discern where the filmmakers got the second half of the unlovely and deeply Silicon Valley coinage “apocaloptimist.” 

The money shot comes from Tristan Harris, cofounder of the nonprofit Center for Humane Technology, who takes in Roher’s point-blank question about whether to have a kid and says the following: “Let’s be honest,” he begins. “I know people who work on AI risk who don’t expect their children to make it to high school.” He holds the off-camera Roher’s gaze for a moment, like the brilliant young oncologist delivering a grim prognosis: “I’m afraid the news isn’t good.” The rest of the film could be filled with sunny, utopian visions without erasing the impact of this moment, and it isn’t.

The real Roher is likely quicker and more perceptive than the querulous 30-something who keeps insisting that these experts rephrase things in plain English and stop freaking him the eff out. The Columbo treatment pays off as he and Tyrell use these expert voices to deliver vivid expository sequences that lay out the basics of AI and the risks it poses. Unfortunately—for him and us—the experts who propose the darkest scenarios appear in the most convincing segment of the film. 

On two facing pages of an open notebook, animated line drawings of Roher and his wife depict a presumably real recorded discussion in which the pregnant wife tells her husband to find some optimism because the stress is bad for their baby. Cue a parade of avuncular middle-aged techno optimists—Peter Diamandis,  Jensen Huang, Reid Hoffman, Guillaume Verdon, and Anthropic co-founder Daniela Amodei—who assure Roher that this is the best possible time to have a child. “Children born today are about to enter a glorious period of transformation,” says the twinkly-eyed Diamandis, founder and chair of something called the Singularity University, who’s among the voices heralding an imminent end to human disease, climate change, energy problems, human labor. As their promises get rosier and more extravagant—with the inevitable reference to populating Mars—you keep waiting for a smash-cut to the black title screen “5 years later,” followed by a slow, wide pan across a post-apocalyptic landscape roamed by unintended consequences. 

Apparently, the filmmakers feel otherwise. They try to resolve the bad news/good news dialectic with celebrity, bringing AI’s bold-faced names to the interview shack so we can meet the Oppenheimers of our moment, to mixed results. While there’s a definite frisson of an intimate audience with people we’ve thus far seen only in clips of Congressional hearings, CEOs like OpenAI’s Sam Altman, Anthropic’s Dario Amodei, and Google DeepMind’s Demis Hassabis are all clearly better at navigating tough questions than Roher is at asking them. (Perhaps sociopathically so, in Altman’s case.) Altman’s opening revelation that he too is expecting a kid earns Roher’s “Mazel tov” and the same softballs pitched to the other star interview subjects, each of whom receive the same softly lit photographic portraiture that made Victor Navalny appear as heroic as he clearly was. They all come off as decent, smart, and sensitive men who clearly want the best for everyone. But that’s hardly the point, is it?

Before landing on a half-hearted call-to-arms—call your Congressmen, educate your friends and children about AI, it’s up to us!—the film makes a glancing mention of the 2023 election in Slovakia, when viral deep-fake audio of the pro-European candidate admitting that he’d rigged the election was released at the perfect time to swing votes to his pro-Putin opponent. But the closest The AI Doc comes to acknowledging the actual imminent threat around a technology as transformative, destabilizing, and power-centralizing as AI is when Tristan Harris says “If we can be the most mature version of ourselves, there might be a way through this.” Take a look at who’s running this country: Is “mature” the word that most readily comes to mind? 

On the other hand, look at Dario Amodei, with his asymmetric smile and affable Wallace-and-Gromit features. Doesn’t he seem like a better person to entrust with this world-historic concentration of wealth and power? Look at the bright-eyed London-accented Demis Hassabis. Shouldn’t we just install him in some governmental role? At the very least, Amodei offers Roher the only worthwhile advice for a father-to-be anxious about the future. “You know, there’s so much uncertainty that I’d just do whatever you were going to do anyway.” From its title, to its graphics, to the phrasing of its questions The AI Doc feels made for children. And, ironically enough, that’s no way to face the future.