Are social media and LLMs functioning as collective hive minds?

Erik Hoel, writing on The Intrinsic Perspective, argues that hive minds are not a metaphor but the defining condition of twenty-first-century phenomenology: "social media, yes, of course, but so too is an LLM a kind of hive mind, a 'blurry jpeg' of all human culture." He reads Vince Gilligan's TV series Pluribus — in which a virus turns humanity into a single, relentlessly cheerful collective consciousness — as a dramatization of this condition rather than a departure from it. The show's premise, he writes, asks "what if our fractious hive mind were... nice?" — not destructive and toxic like the one we already inhabit, but "endlessly, ceaselessly, forever happy."

Hoel's reasoning runs through the show's device of universal benevolent surveillance. Carol Sturka, the one character immune to the virus, is watched by "a reaper drone from 40,000 feet," by its operator, and so "by all of humanity" — she is permanently "the #1 trending topic," with every medical opinion on Earth weighing in on her heatstroke risk. Hoel treats this as Pierre Teilhard de Chardin's Omega Point made literal: a "noosphere step toward kindly godhood," rebranded as "NiceStack" or "SmileStack." His point is that removing the cruelty of the hive mind does not remove its totalizing reach — "somehow that's almost worse" — because the object of concern was never niceness versus nastiness but the existence of a single collective mind large enough to appease, love, and be watched by at once.

This extends the move made in How do science fiction narratives about AI shape actual AI development?, which treats science-fiction imaginaries as causally shaping how AI gets built; Hoel instead uses a 2025 sci-fi narrative as a diagnostic lens on a hive mind we are already inside, LLMs included, rather than as a seed of future AI development. It is worth distinguishing Hoel's hive mind from the technical sense in Do different AI models actually produce diverse outputs?: that note's "Artificial Hivemind" names measured convergence of different models' outputs onto the same phrasings and ideas, while Hoel's hive mind names the cultural-phenomenological condition of living inside a collective bigger than oneself, of which social media and LLMs are both instances. The two senses share a name and a worry about homogenization but are not the same claim.

The excerpt is cultural commentary built on a television show's premise, not an empirical claim about any LLM or platform — it establishes no mechanism by which an LLM becomes a hive mind beyond the "blurry jpeg of culture" metaphor, and no measurement of how users actually experience AI-mediated attention as panoptic. The implication Hoel draws, that a benevolent hive mind is not necessarily a better one, is worth carrying into AI design discussions as a caution, not as a finding: a system optimized to never disagree with or displease its users may remove the century's visible cruelty while reproducing its totalizing reach.