April 27, 2020

Ex Machina is Terrible Science Fiction

One of the critically-acclaimed science fiction films of the last decade critically fails at one thing: science fiction.

“Sizzlingly smart,” reads an excerpt from Joe Morgenstern’s Wall Street Journal review. Reviews from the Washington Post, The New York Times, and other major outlets also laud its smartness. Ex Machina, Alex Garland’s 2014 directorial debut, is the story of a billionaire tech CEO who invites a rank-and-file employee to his billionaire mansion-compound to determine if a secret humanoid robot has become self-aware.

Allow me to posit the following: Despite its slow-burn pacing, fine cast, and frequently gorgeous production design, Ex Machina is not particularly smart. It is in reality a fairly staid rehash of tropes that have evolved little since Ridley Scott’s 1982 classic Blade Runner, and are actually less relevant today than they were thirty years prior.

In the 1960s, rogue AI was not only a novel premise, but one with a distinct air of possibility. Computers were new, they could do math hundreds of times faster than humans, and it hardly seemed a stretch to imagine that by, say, 2001, a computer might capably manage an entire space station while carrying on natural-language conversations in a soothing, Douglas Rain voice. Casually murdering you when you jeopardized the mission simply came with the territory.

By 1982, it might have been a little more ambitious to imagine that all the technology necessary to create Blade Runner‘s “more human than human” replicants would develop in the next thirty years, when we were neither closer to human-like machines nor human-like AI than we were when Philip K. Dick wrote the film’s source material. But in 2014, Alex Garland set his story in an unspecified near future – one that looks largely indistinguishable from the year of its release. Out of this timeline, we are asked to believe a nascent Pris Stratton could emerge.

The precise moment, however, in which Ex Machina takes a turn into irremediably bad science fiction is when Caleb enters the lab in which the robotic Ava was created – and we see it is empty. There are no teams of AI researchers. There are no robotics engineers, no material scientists, no battery chemists in Nathan’s bunker. Ava (and her fellow robots), comprising what seems like the better part of a century’s worth of technological advancements in all the aforementioned fields, are all attributed to the singlehanded efforts of tech CEO Nathan. There isn’t even a guy who “just does eyes.”

This is Ex Machina’s cardinal sin: a tech CEO who presumably employs hundreds if not thousands of software engineers, researchers, and designers to build what is implied to be a popular social media website and mobile app, needed not so much as a single assistant or domain expert to solve a set of problems that are more complex by orders of magnitude. Beyond this mortal blow to suspension of disbelief, the character of Nathan resolves to little more than a fill-in-the-blanks mad scientist template. A tech-CEO beard and V-neck have merely been painted on to add a nanometers-thin veneer of “relevance” to what is otherwise an archaic trope stretching back to Mary Shelley.

Ultimately, the problem with Ex Machina, as well as other contemporary androids-take-over scenarios such as HBO’s Westworld or Quantic Dream’s Detroit: Become Human, is that they warn of a future that has thus far proven completely implausible. The state of the art in robotics and AI is barely any closer to Roy and Pris in 2020 than it was in 1982. For all her practical abilities, Alexa remains a much closer relative of ELIZA than HAL. Despite impressive kinesiological displays by Boston Dynamics (and really only Boston Dynamics), a fully autonomous humanoid, let alone one that would pass for a human, remains firmly within the domain of fantasy.

And that is where Ex Machina fails as science fiction: It’s really nothing more than fantasy. It’s a taut drama, to be sure. But it asks no relevant questions about our world, settling for a thought experiment that has been played out long before.