Joel on Film

The middle section of The AI Doc: or How I Became an Apocaloptimist is the most curious and, likely to some, least useful stretch in Daniel Roher and Charlie Tyrell’s documentary exploring the galvanizing issue of artificial intelligence in our current society. It’s the one in which Roher, who is sort of the co-subject of this movie in addition to its co-director, seeks out a reason to feel hope about the rise and rapid utility of AI from experts who—unlike Roher himself—feel optimism about where things are headed. That’s where the film gets its vaguely Kubrickian subtitle, by the way.

One of the experts who act as Roher’s gateway into this world describes himself as an “apocaloptimist,” a term that he created for himself as a way to reckon with the fact that so many believe AI could lead to something like an apocalypse and this particular subject believing it’ll shape the world for the greater good. Roher likes it and adopts it on the spot as a new term he’ll use. After all, the filmmaker has a good reason to search for the positive side of this issue: He’s about to be a father, and upon feeling a bit of existential dread about his firstborn son growing up in a world partially run by artificial intelligence, he sets out on an ambitious project.

Roher is undoubtedly cynical about all this at the start, but then again, maybe another word for what he is would be “pragmatic.” He has seen, as have we all, the worrisome impact of AI for about the last decade, as well as the leaps in technological efficiency and efficacy that seem to have sprouted up only since the beginning of the recent pandemic. Some of it is good, but most of it, Roher supposes at the start of this even-handed and clear-headed documentary, is probably not so promising, either in its implementation or in the ambitions of those who use it.

Because of his cynicism, which informs a kind of warped sense of humor, Roher has divided the film into chapters—or “parts,” as in “the part where Daniel tries to find hope”—that will explore one particular aspect of this grand conversation before he must move on to the next. The first part is the most depressing one, in which his own and others’ research attempts to explain artificial intelligence beyond the simplistic terms we now know and use. It’s not only limited, for instance, to generating answers or videos in response to questions or prompts, but an actual system, constantly learning and quantifying and, to a frightening degree, coming to conclusions in a scientific context.

That last one is especially alarming to the scientists who have studied the phenomenon of artificial intelligence and, here, are warning against its utility before anything like a generalized study is performed. One visual comparison to fire seems apt: Fire left to its own evolution will destroy, but in hands that have focused it, it’s a tool or weapon. Likewise, the predisposition of AI to learn has its own evolutionary process, but human hands have focused it and, in some cases, aimed it somewhere along the spectrum of human experience.

That second segment, then, sees a number of other experts with a different view of AI as a tool, which is that it can only be used in a way that will, ultimately, lead to good things for the world. There are some truths here, such as how the act of education will change as the reach of knowledge is democratized by access to AI, but those apparent sunny thoughts are followed by a chaser of harsh truth. If unregulated, one scientist says, artificial intelligence might decide to extinguish the human race, and if too heavily regulated, humans with their human aspirations will be in charge of what an all-new intelligence has the ability or permission to say.

That especially is where the film’s third section, which places AI within the context of the current geopolitical climate, comes in, because the issue of regulation has long since died on the vine. World leaders leapt into action to decide, separately from each other, how far they’re willing to push AI as both a tool and a weapon. Tech companies, represented here in interviews with Sam Altman (of OpenAI), Dario Amodei (of Anthropic) and Demis Hassabis (of Google DeepMind), put this untested and sort of unregulatable intelligence in a competitive market between each other (along with the representatives of Meta and xAI who, unsurprisingly, do not show up in this movie).

Those interviews do very little either to dissuade Roher’s fears from the beginning of the movie or to undergird his later-found optimism, but that’s also likely because we’re a good decade away from really understanding AI as more than a sentience and a tool. What happens within that decade, the film charges, is at the whim of a number of factors that are simply unknown and uncontrollable. In a telling moment, Altman confirms that it would be futile for Roher to ask him to make any promises about a positive outcome to all this.

There might be hope, of course, but expecting it to arrive at the middle portion of this movie is about as foolish as expecting this movie to be the definitive documentary about artificial intelligence at this current moment. Despite its surtitle, The AI Doc: or How I Became an Apocaloptimist is not—could not be—that movie, but it is a good and informative film all the same, alternately sober and, because it simply must be, playful.

Rating: *** (out of ****)