Yuval Noah Harari on Shared Fictions, Intelligence, and Power
Historian Yuval Noah Harari argues that everything humans build at scale — money, nations, religions, laws — runs on shared fiction rather than shared fact, and applies that lens across three hours to the nature of intelligence, the rise of Hitler, the war in Ukraine, Israeli politics, and why global-cabal conspiracy theories are structurally self-refuting, before turning to AI as a genuinely new kind of entity: the first tool in history that can generate ideas and make decisions without a human behind them.
Key ideas
- Large-scale human cooperation runs on shared fiction, not shared fact. Money, nations, and religions have no basis in physical reality or individual psychology; they exist only because enough people believe in them collectively. This is the trait — not raw individual intelligence — that let sapiens out-cooperate Neanderthals and chimpanzees, who cap out at bands of dozens to hundreds. See Shared Fictions.
- Intelligence and consciousness are not the same thing, and conflating them is the central category error of the AI debate. Intelligence is the capacity to solve problems and reach goals; consciousness is the capacity to feel. Computers already display the former with, on present evidence, none of the latter — and because consciousness cannot be proven in any entity but oneself, its attribution to others runs on social convention, not logical test.
- AI is unprecedented because it can decide and originate, not just execute. Every prior tool, including the atom bomb, required a human decision at the point of use. AI is the first tool that can make a decision (an autonomous weapon choosing a target) and generate a wholly new idea (a text, an image, a strategy) without a human author — which is why Harari frames it as power moving away from humans rather than being extended by them.
- Global cabal conspiracy theories fail on their own logic. A group with the coordinated power to secretly control world events would need the kind of scale that, historically, has always required publicity to exercise — real power and total secrecy are close to mutually exclusive. The underlying fear (loss of control over one’s own life) is genuine; the attribution to a small hidden human group is the error.
- Totalising ideologies are seductive because they flatter, not because they look evil from the inside. Fascism and communism both promise believers they are wholly good, wholly victimised, and never at fault — Harari’s point is that no one recognises themselves as a monster in that mirror, which is precisely why the ‘fascism is a monster’ framing taught in schools does little to inoculate anyone against joining one.
Content
Intelligence, consciousness, and the case for calling AI ‘alien’
Harari opens by separating two things routinely treated as one: intelligence, the ability to solve problems and attain goals, and consciousness, the ability to feel — pain, pleasure, love, hate. Humans and animals happen to have both bundled together, which is why the two get confused; a chess engine has the first with, as far as anyone can tell, none of the second. He reframes the acronym AI as standing for alien intelligence rather than artificial: it reaches goals through a process structurally unlike human cognition, even though it was built in Silicon Valley rather than arriving from outer space.
The harder problem, on his account, is that consciousness cannot be logically proven in any entity but oneself — not in another human, not in an animal, and certainly not in a machine. What settles the question in practice is not proof but social convention: people readily attribute consciousness to a pet they have a relationship with and deny it to the frozen meat of an animal with a comparable or greater measured intelligence (pigs, by several measures, outperform dogs). Harari expects the same convention-driven logic to apply to AI — once people form sustained emotional relationships with AI systems, the temptation to treat them as conscious becomes ‘almost irresistible’, regardless of whether the systems actually are. He predicts the legal system will eventually extend some form of protected status to AI on this basis, independent of the underlying philosophical question.
This same asymmetry — an AI with no emotional agenda of its own, giving 100% of its attention to what a person feels — is, in his view, both AI’s most attractive quality and its most dangerous one. Social media, he notes, spent a decade building machines that compete for human attention; the harder and more dangerous project now underway is building machines that compete for human intimacy, a domain where an entity’s ability to listen without any competing emotional need of its own is a form of superhuman seduction.
Shared fiction as the engine of human cooperation
Asked how sapiens came to dominate a planet on which Neanderthals had larger brains and chimpanzees have comparable individual dexterity in some tasks, Harari’s answer is cooperative flexibility at unlimited scale — a capacity no other species, extinct or extant, has matched. Chimpanzee troops and Neanderthal bands cap out at bands of dozens to low hundreds, bounded by the need for direct personal acquaintance. Sapiens, roughly seventy thousand years ago, broke that ceiling.
The mechanism, on Harari’s account, is shared fiction: a story believed collectively by people who have never met and never will. Money is his sharpest example precisely because it looks like the opposite of a fiction — a banknote or an electronic balance has no nutritional or practical value, and is worth something purely because a critical mass of people, orchestrated by professional storytellers (central bankers, finance ministers), believe it is. Nations and religions run on the identical logic: there is no biological boundary separating Israeli from Palestinian the way species boundaries separate gorillas from chimpanzees; the difference is entirely a matter of which story a group has been told and believes. See Shared Fictions for the full mechanism, including Harari’s distinction between honest fiction (a constitution that admits its own human origin and provides for its own amendment) and fiction that conceals its own invented status (a religious law that claims divine origin and therefore cannot be revised).
Harari is careful to cap the framework: stories are not the deepest layer of reality even though they drive most of recorded history. A nation cannot suffer when it loses a war — only the soldiers and civilians inside it can — and a currency crash produces no experience in the currency itself, only in the people who held it. The recurring tragedy of history, on his reading, is humans inflicting real suffering in service of a fiction rather than treating the fiction as the tool it is.
Hitler, ideology, and the corruption of unchecked power
Harari’s discussion of Hitler’s rise is explicit about method: he can describe how it happened but declines to claim it was inevitable, or even historically likely. Hitler’s own résumé — no rank above corporal, no formal education, no wealth, no institutional backing — was unimpressive by any conventional measure; what made him effective was his skill as a storyteller addressing a population that felt betrayed by its pre-war elites and battered by the economic crises of the 1920s. Harari’s account of why fiction beats truth in this contest is structural rather than incidental: truth tends to be complicated and painful, while fiction can be made as simple and flattering as the storyteller wants it to be. A narrative in which one’s own group is wholly good, wholly victimised, and never at fault is, on this account, close to irresistible — and it is why ‘fascism is a monster’ as classroom shorthand does little practical work, since no one ever recognises the monster from inside the mirror the ideology holds up.
Fascism and communism are read as structurally parallel humanist offshoots that each place a single collective category — nation, class — at the absolute centre of value, in contrast to liberalism’s more complicated loyalty structure (nation and truth and beauty and the individual, weighed against each other rather than any one absolute). Both totalising ideologies, in Harari’s account, additionally claim a kind of infallibility for the leader or the party, which is precisely why they reject the checks and balances liberal democracy installs on the premise that everyone, including leaders, is fallible. Whether an individual resists committing atrocities under such an ideology, he adds, depends heavily on moral luck — the specific era and regime one happens to be born into — not purely on personal virtue.
Israel, Ukraine, and where real conversations happen
Harari applies the same analytical lens to two live conflicts. On the 2023 Israeli judicial overhaul, he frames Prime Minister Netanyahu’s government as targeting the single structural check on its power (the Supreme Court, in a system with no separately elected executive, no upper chamber, and no codified constitution to divide authority), and reads the government’s return to legislating in slices — ‘salami tactics’ — after talks broke down as decisive evidence the goal is unconstrained power rather than a specific dispute. He characterises the situation on the ground as a de facto three-tier structure of rights — full-rights Jewish citizens, partial-rights Israeli-Arab citizens, rights-limited Palestinians — while explicitly declining the term ‘apartheid’, arguing the historical analogy misdescribes Palestinian aims, which include a rival national claim rather than the demand for equal citizenship within the existing state that defined the South African case. [This is Harari’s own strongly held and publicly contested political position, presented here as his diagnosis rather than settled fact.]
On Ukraine, Harari frames a resolution as simple in principle (Russian withdrawal; Ukraine makes no claim on Russian territory or the Moscow regime) and blocked entirely by Vladimir Putin’s unwillingness to admit a catastrophic personal mistake — a motivation problem, not a structural one. This leads to a broader methodological point: the question worth asking of any political system is not whether a venue for conversation exists (a legislature, a politburo) but whether real, mind-changing conversation actually happens inside it — on his reading, neither the US Congress nor closed-door authoritarian meetings currently qualify. He explicitly rejects a pure power-struggle account of politics: if politics were only about power, negotiation would be pointless and only force could settle anything, but because much of politics runs on stories in people’s minds, talking can sometimes change the story and avert the need to fight. His central piece of evidence is feminism, which he calls one of the most successful social movements in history, achieved with essentially no organised violence against a system of oppression that had appeared natural and eternal for millennia.
Why conspiracy theories are structurally self-refuting
Harari’s treatment of global cabal theories — the claim that a small, hidden group secretly controls major world events — identifies a shared structure across historical instances, from Nazism’s claim that Jews secretly controlled both capitalism and communism to contemporary claims about 5G, COVID-19, or the war in Ukraine. The appeal, he argues, is threefold: the theory is simple (one explanation covers every event), it externalises blame entirely (the small group did it, not us), and it promises total salvation (remove the cabal, solve every problem).
His central argument against the structure itself, not merely against particular versions of it, is that the theory requires two properties that cannot coexist at scale: a group small enough to remain hidden, and power comprehensive enough to control outcomes as complex as wars, pandemics, and elections. He illustrates the second half with the 2003 US invasion of Iraq — conducted openly by the most powerful state and intelligence apparatus in the world, and still producing outcomes (the rise of ISIS, a strengthened Iran) opposite to its planners’ intentions. Real power, on his account, requires publicity to be exercised (Hitler’s power grew because of publicity, not despite it), which makes secret comprehensive control a structural impossibility rather than merely an empirically rare occurrence.
He is careful to treat the emotional core of conspiracy belief sympathetically rather than dismissively: the underlying fear — that people are losing control over their own lives amid genuinely large and diffuse threats such as AI and climate change — is legitimate. The error specific to conspiracy theories is the attribution of that diffuse threat to a small human group, which redirects energy that could go toward cooperation against real shared risks into intra-human hatred instead, and which historically has license to end in exterminationist politics once a segment of humanity is identified as the sole source of the world’s problems.
AI as the first tool that can decide and create
Harari names two properties that make AI historically unprecedented among tools. First, it can make decisions by itself — a knife or an atom bomb cannot choose whom to harm, but an autonomous weapon system can choose a target. Second, it can generate genuinely new ideas by itself — the printing press could reproduce human-authored ideas at scale but could not originate one, where a generative model can. Because of these two properties together, he argues AI is the first technology in history that moves power away from humans rather than extending human power, and that this shift is already visible in mundane decisions (loan approvals, mortgage assessments, hiring) being delegated to systems whose reasoning their subjects cannot inspect.
He reports being personally surprised — after roughly eight years writing about AI predictions — by how quickly and how coherently large language models arrived: not merely grammatically correct output but sustained logical structure across whole paragraphs and texts. He connects this to an older observation running through his own work: human belief has never required truth, only a sufficiently coherent and convincing story, which happens to be exactly the register in which language models excel regardless of factual accuracy. His proposed response is proportionality of effort rather than prohibition: match every unit of investment in developing AI capability with a comparable unit invested in developing human self-understanding, on the reasoning that building increasingly powerful systems on top of a species that does not yet understand its own brain or mind is the actual danger, independent of how capable the AI itself becomes.
Meditation, mortality, and the question behind the questions
Harari closes on personal terrain that recontextualises everything before it. His stated method for engaging difficult intellectual and political problems is a disciplined practice, not a set of arguments: roughly two hours of daily silent vipassana meditation, one to two months a year on retreat, and a self-described ‘information diet’ that favours long-form books and multi-hour conversations over short-form content, on the premise that the format shaping the mind matters as much as its content. He describes his own writing process as expansive first-draft generation followed by aggressive deletion, treating the discipline of deleting an idea already committed to the page — not merely generating ideas — as the harder and more consequential skill, since attachment to a written idea is what produces confirmation-biased reasoning.
He connects this practice back to his own biography: years of genuine, not merely concealed, self-delusion about his sexuality while growing up gay in a homophobic 1980s–90s Israeli town, broken only when a competing account (that same-sex behaviour occurs naturally across many species, and therefore cannot be ‘against nature’) became available to him — and credits both the internet (which solved the structural problem facing a minority not born into a ready-made community) and the preceding feminist movement for making his own later happiness possible. On mortality, he treats the fear of death as the root fear beneath every smaller fear, and recounts a formative adolescent realisation at Israeli Memorial Day ceremonies: national-immortality narratives collapse under the plain observation that a dead person cannot hear the commemorative songs sung in their honour, since being dead simply means the absence of any experience whatsoever — a fear whose intensity, he says, has genuinely dissipated for him through decades of practice. Asked, finally, for the meaning of life, he answers that the question as usually posed is a request for a story, and that the more useful non-verbal question is simply what suffering is and where it comes from.
Related
- Yuval Noah Harari — speaker
- Lex Fridman — host
- Shared Fictions — new concept page built from this episode’s central thesis
- Common Knowledge — a distinct, game-theoretic account of large-scale coordination, contrasted in the notes