How Pluribus nails AI’s great flattening effect
What lessons does the hivemind in the drama Pluribus have for humanity in thrall to artificial intelligence?
Do individuals matter more than the collective? Is a world where every human contentedly works towards common goals better than one in which selfish motives constantly clash? Is human creativity, born of individual expression, worth the price we pay for it in violence? These are some of the questions Pluribus, the new Apple TV+ drama by Vince Gilligan, creator of Breaking Bad and Better Call Saul, asks without undue subtlety. We are just five episodes in, but they are not the subtext of the show—they are the show.
Spoilers follow, if you care about such things and have not tuned into any of the discourse around the series. In Pluribus, an alien virus takes over humanity and turns people into zombies—only, these are the nicest, kindest zombies ever seen in fiction. All humans on earth—barring a handful that are immune to the virus—become part of a hivemind that rejects violence and lives in a state of apparent harmony and cooperation. The protagonist of the show, one of the immune, is Carol Sturka, an author who made a fortune writing popular bilge while secretly longing to be taken seriously. Fiercely independent, she is constantly at odds with the hivemind, rejecting the idea of a common utopia and railing against the loss of individual will that it represents.
It remains to be seen what their ultimate goals are, but for the time being, the “us" in Pluribus are happy to run the world with calm efficiency and order. It is fascinating to see what happens when the entirety of human knowledge is shared—everyone can do everything in this world, from flying planes to conducting complex surgeries. The sum total of human knowledge now resides in every human brain.
If that reminds you of generative AI, that is most likely as Gilligan intended. Not only has the screenwriter-director been quite open in his criticism of AI, calling it a “plagiarism machine", the credits of the show actually contain the line “this show was made by humans." In multiple interviews, he has expressed his scorn for AI and its shiny, regurgitated cleverness, and it is natural to assume that the wholesome and all-knowing hivemind in Pluribus is a stand-in.
Turns out, human researchers were already on to the idea of AI being essentially a hivemind that pretends to think independently but produces results that it estimates would be most palatable. An award-winning paper Artificial Hivemind: The Open-Ended Homogeneity of Language Models (and Beyond) published in October by NeurIPS, a conference on neural information processing systems, explores the idea by putting LLMs through common queries, to which they all deliver standardized responses. “Language models (LMs) often struggle to generate diverse, human-like creative content, raising concerns about the long-term homogenization of human thought through repeated exposure to similar outputs," the authors of the paper write.
What does generative AI do, after all, except take the sum total of human knowledge, acquired by people working through human constraints—pain, grief, heartbreak, crippling procrastination—and make it all look easy? It sands away at the edges of human creativity and expression—crafting a sentence, writing dialogue, making sense of the world through words wrenched out of one—and turns it into a bland alphabet soup that seemingly makes sense, sure, but is devoid of any taste and flavour. AI does not challenge, it does not doubt, and it certainly does not create. It is the snake eating its own tail, and as Pluribus warns, it erodes individuality, the very thing that makes us create in new ways.
topics
