The Fiction Monopoly

In the AI Story

Harari's framework locates the origin of human dominance in the Cognitive Revolution—a neurological shift occurring between seventy and thirty thousand years ago that gave the species symbolic thought. This was not a bigger brain or better tools; it was the capacity to construct and inhabit shared fictions. A Sumerian merchant and priest who never met could build a ziggподатака because both believed in the same gods. Two twenty-first-century strangers execute financial transactions because both believe in banks, contracts, money—none of which exist outside collective agreement. The fictions are not decorations on material reality. They are load-bearing walls. Remove the shared fiction of the United States, and what remains is not a simpler polity but three hundred thirty million primates without coordination infrastructure beyond personal acquaintance.

Every previous amplification technology—printing press, radio, television, social media—expanded the reach of human storytelling without displacing the storyteller's role. A human decided what to print, broadcast, tweet. The technology magnified human narrative capacity; it did not replace the human narrator. AI crosses a categorical boundary: it generates narrative without a narrator. When a large language model produces a paragraph about justice, the text carries the full intersubjective weight that 'justice' has accumulated across centuries—legal precedents, philosophical debates, moral intuitions. The surface is indistinguishable from human-authored prose. The interior is empty. No belief. No stakes. No participation in the community whose shared meanings the word carries.

The monopoly broke not when AI became conscious—it didn't—but when it became fluent. The winter of 2025 marked the threshold: systems crossing into practical indistinguishability from human-generated text for most readers in most contexts. This is not hyperbole. Harari's claim in Nexus that 'AI has hacked the operating system of human civilization' encodes structural precision. If civilization runs on shared fictions, and if the production of those fictions is no longer exclusively human, then the coordination mechanism that built every institution—from marriage to the Geneva Conventions—has been fundamentally altered. The implications emerge when you map this against catastrophic risk: counterfeiting. One fake bill in a stack is harmless. As the ratio shifts, the currency's reliability degrades. AI-generated text is not legal counterfeit, but it operates by identical logic—mimicking genuine intersubjective participation with sufficient precision that the distinction becomes invisible.

Origin

Harari developed the fiction-cooperation thesis across Sapiens (2011), synthesizing evolutionary biology, cognitive archaeology, and comparative anthropology. The foundational claim—that large-scale human cooperation depends on belief in entities that exist only in collective imagination—was not original to Harari. Elements appear in Émile Durkheim's collective conscience, in Benedict Anderson's imagined communities, in the symbolic interactionism of George Herbert Mead. Harari's synthesis organized these scattered insights into a single historical narrative spanning from Paleolithic bands to global capitalism.

The monopoly concept gained urgency in Homo Deus (2015) and became alarm in Nexus (2024) as large language models crossed capability thresholds Harari had assumed remained distant. His 2023 Economist essay—'AI has hacked the operating system of human civilization'—marks the moment his framework shifted from historical analysis to immediate warning. The hacking metaphor is deliberate: an operating system runs invisibly, governing everything; language is the OS on which human civilization runs; AI has achieved write-access.

Key Ideas

Fictions enable strangers to cooperate. No other mechanism—kinship, personal acquaintance, contractual enforcement—scales to billions. Money, nations, human rights, corporations exist only as collective beliefs, yet they coordinate behavior more reliably than physical coercion.

The monopoly was complete and invisible. For seventy thousand years, every shared fiction originated in human minds with stakes in the world those fictions governed. The storyteller participated in the story.

AI breaks the monopoly without becoming conscious. The threshold is fluency, not sentience. Systems can now generate intersubjective content—narrative that reads as though a believing mind produced it—without belief, understanding, or participation.

The break is a phase transition, not incremental change. Like water to ice: same substance, sudden reorganization. The winter of 2025 was not a gradual improvement in text generation; it was a categorical crossing into practical indistinguishability.

Counterfeiting is the historical parallel. One fake bill is harmless; as the ratio shifts, currency reliability degrades. AI-generated text mimics genuine intersubjective participation; as the proportion of such text rises, the trust infrastructure of civilization erodes.

Debates & Critiques

Critics challenge whether current AI systems truly 'generate fictions' or merely recombine training-data patterns—a distinction Harari's framework treats as less important than the functional reality that the outputs coordinate behavior regardless of the generation mechanism. Others argue the monopoly was never complete—that institutions, not individuals, have long produced the fictions (legal codes, corporate charters) and that AI simply automates institutional production. The rejoinder: institutions were staffed by conscious humans with stakes; AI systems are not.