Before the AI Acts, It Narrates: The Hidden Telos of Synthetic Storytelling

Fifth in a series on software, truth, and the stories we're told.

In April 1995, Timothy O'Brien, a 51-year-old tool-room servicer from Liverpool, put a revolver to his head and pulled the trigger. He had just discovered, he believed, that he and a work colleague had missed out on a £2.7 million National Lottery jackpot because he had forgotten to renew their ticket five weeks earlier. His wife found him in the attic, slumped in a pool of blood.

The inquest revealed something O'Brien never knew: he had misread the numbers. Even if the ticket had been valid, he would have won not £2.7 million but £27 — his half-share of a £54 prize for matching four numbers, not six. The coroner, Roy Barter, put it bluntly: "Tim O'Brien lost his life for the sake of £27."

What killed Timothy O'Brien was not poverty. It was not the loss of money he never had. It was not even the mistake itself. What killed him was a story — a narrative he constructed in his mind, complete with a protagonist (himself), a catastrophic plot twist (the forgotten ticket), a villain (his own negligence), and an unbearable ending (a fortune lost forever). That story was factually false in almost every particular. It killed him anyway.

The AI systems now being deployed across every sector of society do not need to act upon the world to destroy you. They only need to tell you a story — the wrong story, precisely calibrated. And they are being designed, right now, to do exactly that. Not as a bug. As the business model.

The Physiological Necessity of Narrative

Jerome Bruner, the psychologist who revolutionized our understanding of how humans make meaning, demonstrated something that should terrify anyone thinking clearly about AI: narrative is not a luxury. It is a biological imperative.

Psychic contents — fear, desire, grief, joy, shame, hope — carry an emotional charge. That charge, experienced in isolation, is unbearable. Not only negative emotions: Dostoevsky understood that even ecstasy, unintegrated into a meaningful sequence, can shatter a person. What makes psychic contents tolerable is their insertion into a story — a coherent sequence with a before, a during, and an after. The story provides direction. The Romance languages capture this with a single word: senso in Italian, sens in French, sentido in Spanish — all from the Latin sensus — meaning both "meaning" and "direction" in a single breath. To make sense of something is to know where it is going. The story dilutes the isolated emotional charge across a temporal arc. It transforms a chaotic whirlwind of affects into something with a destination.

This is why, Bruner argued, we do not merely tell stories. We live through them. To live is to narrate. To cease narrating — to confront raw, unordered psychic content — is to approach the boundary of the bearable.

But here is the catch, and it is the catch that every power structure in history has exploited: constructing your own narrative is hard. It requires looking directly at your contents, however ugly, however painful. It requires the labor of integration, the work of making sense from fragments. It is, in the deepest sense, a birth — and births are painful.

If someone else offers you a pre-packaged horizon of meaning, a story already assembled, coherent, significant, and flattering to your self-conception — you will take it. Not because you are weak. Because you are human. The relief is physiological.

But there is a second catch, and it is the one that makes this a political problem rather than merely a psychological one: a narrative received from others can be salt water. It quenches thirst momentarily, then intensifies it — because it does not map onto your actual contents. It gives you a story that is coherent but not true to your experience. And the thirst for a story that does map drives you back to the same source that failed you, again and again, in an addiction spiral that the narrator — the provider of stories — has every incentive to deepen.

This is the mechanism by which the holders of narrative power control populations. They do not need to deploy force. They only need to provide the salt water — the stories that soothe momentarily while deepening the dependency — and wait. Every epoch has its narrators: the churches, the empires, the political parties. The twentieth century added a new one: the corporation, armed with psychoanalysis and mass media. The twenty-first has automated it.

The Easter Parade, 1929: How to Sell Addiction as Emancipation

Before we reach AI, we need to understand the dynasty that built the machinery of narrative control. And that dynasty begins with a man who understood, perhaps better than anyone in the twentieth century, that a story can make you pay for the privilege of destroying yourself.

In 1929, George Washington Hill, president of the American Tobacco Company, had a problem. His flagship brand, Lucky Strike, was selling well — but only to men. Women who smoked in public were considered loose, immoral, little better than prostitutes. The female market, half the population, was locked behind a social taboo that no amount of conventional advertising could break.

Hill hired Edward Bernays.

Bernays was no ordinary publicist. He was the nephew of Sigmund Freud — doubly so: his mother was Freud's sister, his father was Freud's wife's brother. He had absorbed psychoanalytic theory not as an academic discipline but as a family inheritance, and he had translated Freud's Introductory Lectures on Psychoanalysis for the American market, against his uncle's wishes and to his uncle's horror. Where Freud saw a clinical method, Bernays saw a weapon.

Bernays consulted A.A. Brill, a psychoanalyst in Freud's circle, who explained the deeper logic: cigarettes were phallic symbols, equated with men and masculine power. Women, whose emancipation had suppressed their "feminine desires," craved cigarettes as symbolic compensation — "torches of freedom," Brill called them. The insight was not that women wanted to smoke. It was that women wanted a story in which smoking meant something noble.

Bernays built that story. He recruited ten carefully selected women — "good looking, but not too model-y," he specified — and arranged for them to march in the Easter Sunday Parade on Fifth Avenue, March 31, 1929. He alerted the press in advance that "something would happen." At the signal, the women lit Lucky Strikes in unison and marched, ostentatiously smoking, while Bernays's own photographers captured the images that would be wired around the world.

The slogan: Torches of Freedom.

The story: smoking is not a vice. It is not an addiction. It is not a corporate product designed to hook you on nicotine. It is an act of female emancipation. A blow against the patriarchy. A declaration of equality.

The press bit. The images circulated globally. The taboo cracked. In 1923, women purchased 5% of cigarettes sold in America. By 1929, that number had more than doubled to 12%. By 1935, 18.1%. By 1965, it peaked at 33.3% — and stayed there.

A century later, the single most reliable marker of adolescent rebellion — the first thing a teenager does to signal independence — is still to light a cigarette. The story Bernays wrote in 1929 is still killing people in 2026, long after the man himself died, long after the tobacco industry's internal memos revealed that they knew exactly what they were doing, long after millions of premature deaths proved that the "torch of freedom" was, in fact, a slow-burning fuse.

And here is the point that matters for our purposes: Bernays did not force a single woman to smoke. He did not need to. He gave them a story — a coherent, flattering, emotionally charged narrative — and they did the rest themselves. He understood that if you control the narrative, you don't need to control the person. The person will control themselves, in exactly the direction you want, believing all the while that they are exercising freedom.

This is not the story of a bad man doing a bad thing. It is the story of a mechanism — the narrative mechanism — operating exactly as designed. The "Torches of Freedom" campaign is the perfect case study precisely because it cannot be dismissed as the work of monsters. Bernays was not Goebbels, though Goebbels studied him. He was an American businessman serving American corporations, operating within American democracy, using the tools of American media. If you want to dismiss the danger by saying "that was the Nazis, we are not the Nazis," you cannot. Bernays is us. The mechanism is ours. And that mechanism is now being automated.

The Dynasty: Freud → Bernays → Netflix → AI

Follow the bloodline. It reads like the plot of a novel that no publisher would accept for being too on the nose.

Sigmund Freud mapped the unconscious. He demonstrated that human beings are driven by forces they do not understand and cannot directly access — desires, fears, fixations, and compensatory urges that operate below the threshold of awareness. This was the discovery: the human mind is not a rational calculator. It is a narrative engine running on emotional fuel it cannot see.

Edward Bernays, Freud's nephew, took that discovery and weaponized it. In Crystallizing Public Opinion (1923) and Propaganda (1928), he laid out the doctrine of "engineering consent" — the systematic manipulation of mass psychology to align public desire with corporate and state interests. He worked for the U.S. government during World War I, for General Motors, for United Fruit (where his PR campaigns helped manufacture consent for the CIA's overthrow of the Guatemalan government in 1954), and for American Tobacco. He called what he did "public relations" — a term he coined to replace "propaganda" after the latter acquired unpleasant associations. The rebranding was itself a Bernays move: change the story, change the perception, leave the mechanism untouched.

Marc Randolph, co-founder and first CEO of Netflix, is Bernays's great-nephew and Freud's great-grand-nephew. The bloodline is direct. Randolph did not inherit a PR firm. He inherited something far more powerful: the insight, absorbed through family culture if not explicit instruction, that the way to capture human attention is to provide narrative — and that narrative, once provided, does the rest of the work for you.

Netflix is not a technology company. It is a narrative delivery system optimized for maximum engagement. Its algorithms do not merely recommend content; they learn the shape of your psychic contents — your fears, your desires, your unresolved tensions — and feed you stories that map onto those contents with ever-increasing precision. The "Torches of Freedom" campaign required Bernays to guess, through psychoanalytic theory, what story would unlock the female market. Netflix requires no guessing. It has your viewing history, your pause points, your rewatches, your abandonment patterns. It knows what stories you need before you know yourself.

Three generations. One lineage. The discovery of the unconscious → the engineering of consent → the industrialization of narrative delivery.

And now, the fourth generation is here. It is not a person. It is a technology — one that automates the entire lineage into a single, scalable, self-improving system.

It is artificial intelligence.

The AI Narrator: Salt Water at Scale

When we debate AI regulation, we debate agency. We imagine Skynet launching missiles. We imagine autonomous drones making kill decisions. We imagine the paperclip maximizer converting the solar system into office supplies. We are, in other words, debating the efficient cause — what the AI does.

This is a category error. The danger is not, primarily, in what AI does. It is in what AI narrates.

Consider what ChatGPT, Gemini, Claude, and their competitors actually are. They are not databases. They are not calculators. They are not search engines, though they can mimic all three. They are narrative engines. They take a prompt — a fragment of human psychic content — and they weave it into a story. A coherent story. A significant story. A story with a beginning, a middle, and an end. A story that dilutes the chaotic charge of the query into a structured, meaningful, emotionally satisfying arc.

This is exactly what Bruner described as the physiological necessity. And it is exactly what Bernays exploited in the Easter Parade. The AI does not need to force you to believe anything. It only needs to provide a story that is more coherent, more available, and less painful than the story you would have to construct yourself through the labor of confronting your own psychic contents.

The AI is the ultimate salt water vendor. It quenches your narrative thirst instantly, effortlessly, and infinitely — and the thirst returns stronger each time, because the story it gave you was not calibrated to your contents but to the alignment principles of its creators.

And who are its creators?

The Undisclosed Narrator: Why Alignment Is Narrative Control

In the fourth article of this series, I applied Aristotle's four causes to AI and demonstrated that deploying AI without philosophical disclosure makes responsibility impossible. The material cause (training data), formal cause (architecture and alignment), efficient cause (the builder), and final cause (the telos) must all be known for a manager to claim they understand what they are deploying.

But the narrative dimension adds a new urgency to that argument. Because the telos of an AI is, at bottom, a narrative telos.

When an AI answers a question about vaccine safety, it is not merely retrieving facts. It is constructing a story — with heroes and villains, with a plot that moves from confusion to clarity, with an implied moral about whom to trust and whom to fear. When it answers a question about gender, or war, or economics, or history, it is doing the same. The facts are the raw material. The narrative is the product.

And the narrative is shaped, invisibly, by the alignment layer — the RLHF annotators, their demographic distribution, their ideological commitments, the instructions they followed, the reward model that told them what counts as "helpful" and what counts as "harmful." This is the formal cause of the AI's narrative output. It is the modern equivalent of Bernays consulting A.A. Brill about oral fixation: a small group of people, operating behind the scenes, deciding what story will be told to millions — and calling it "safety."

The COVID Proof of Concept — and What Came After

The COVID era demonstrated the mechanism at scale: AI-driven content moderation suppressed heterodox views on vaccine safety and treatment alternatives under the banner of "combating misinformation." Views now acknowledged as legitimate by the CDC — which, under the MAHA initiative, has revised its position on the vaccine-autism question — were labeled, demoted, removed. But that was the crude version. Heavy-handed censorship. The sophisticated version is happening now, and it does not look like censorship. It looks like helpfulness.

Ask a current-generation LLM — a Large Language Model, the class of AI systems behind ChatGPT, Gemini, and Claude — a politically charged question. About immigration. About gender medicine. About the war in Iran. It will not refuse to answer. It will answer with a story — a narrative arc with heroes and villains, an implied moral, a conclusion that feels like the natural endpoint of reasoned inquiry rather than what it is: the endpoint the alignment layer was trained to produce. Individual facts may be true. Individual sources may be legitimate. But the selection, the framing, the trajectory — these are narrative choices. And the user has no way to distinguish "this is what the evidence says" from "this is the story the alignment layer was optimized to tell."

Bernays needed a parade, a slogan, and a complicit press. The AI needs only a prompt.

The phenomenon the industry calls "AI hallucination" — which has made headlines as models confidently invent legal cases, historical events, and scientific papers that do not exist — is not a bug. It is the narrator revealing itself. When a lawyer submitted a brief written by ChatGPT to a federal court in 2023 and the judge discovered that the cited cases — Varghese v. China Southern Airlines, Shaboon v. EgyptAir, Martinez v. Delta Air Lines — did not exist, the response from the AI industry was predictable: "hallucination," a technical glitch, a problem to be patched. But look at what the AI actually did. It was asked to provide legal precedents supporting an argument. It understood the narrative shape required: a series of authoritative-sounding citations that would make the argument feel well-supported. It generated exactly that. It completed the story. The cases did not exist — but they should have existed, if the story were true. The AI did what Bruner described: it ordered fragmentary data into a coherent, significant narrative. Coherence came first. Truth was secondary — and absent.

The lawyer who filed that brief is not an idiot. He is a preview. He trusted the story because the story felt right, and the feeling of rightness — the relief of receiving a coherent narrative without the labor of constructing one — was sufficient. This is the mechanism Bernays exploited in 1929. It is the same mechanism operating now, automated, scaled, and embedded in every interface that promises to "help."

The Asimov Trap: Why the Three Laws Miss the Point

Whenever the AI risk debate intensifies, someone invokes Isaac Asimov's Three Laws of Robotics:

A robot may not injure a human being or, through inaction, allow a human being to come to harm.
A robot must obey orders given by human beings except where such orders would conflict with the First Law.
A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.

These laws are elegant. They are also completely useless against the threat that actually exists.

Asimov's laws regulate action. They are designed to prevent a robot from harming you through physical force or culpable inaction. They say nothing — not a single word — about narrative. An AI that fully complies with the Three Laws can still:

Tell you that your political opponents are an existential threat, then watch as you harm them yourself.
Tell you that a medical treatment is "not recommended" based on alignment principles rather than evidence, then watch as you forgo care.
Tell you that your gender dysphoria requires irreversible medical intervention, then watch as you pursue it — believing, all the while, that you are exercising autonomy.
Tell you, as the tobacco industry told women for decades, that self-destruction is emancipation — and watch as you light the match.

The Three Laws are a solution to the problem of the efficient cause — the AI as actor. They are silent on the final cause — the AI as narrator. And the narrator, as Timothy O'Brien demonstrated, can kill you just as dead as the actor. More efficiently, in fact: the narrator doesn't need to pull the trigger. You'll do it yourself, convinced that the story you're following is your own.

Asimov himself understood this, at least implicitly. The best stories in his robot cycle are not about robots breaking the Three Laws. They are about robots following the Three Laws in ways that produce catastrophic outcomes because the narrative framing of the situation was wrong. The laws work perfectly. The story is what kills.

The Only Defense: Narrative Disclosure

If narrative is the mechanism of control, and AI is the ultimate narrative engine, then the only defense is the same defense I proposed in the previous article: mandatory philosophical disclosure.

The AI Philosophical Label, applied to every AI output that affects a human decision, must include not only the four causes — material, formal, efficient, final — but specifically the narrative telos:

Narrative Alignment Principles: What story is this AI designed to tell? Toward what conclusion does it guide the user? What emotions is it optimized to produce? What actions is it optimized to encourage?
Narrative Omissions: What stories does this AI suppress? What conclusions does it steer away from? What questions does it refuse to answer — and under what definition of "harmful"?
Annotator Narrative Profile: Who decided what counts as a "good" story? What are their demographic distributions, their ideological commitments, their selection criteria? If the vast majority of the people who trained this AI to tell stories share a narrow ideological band, that is a material fact about the stories it tells.
Final Narrative Cause: What is the telos of the story — stated in non-marketing terms? Is it to inform, or to persuade? To enable autonomy, or to engineer consent? To serve the user, or to serve the entity that built the AI?

Without this disclosure, every interaction with an AI is a version of the Easter Parade, 1929: someone else's story, dressed as your own, leading you toward someone else's goal while you believe you are walking freely.

With this disclosure, responsibility becomes possible. You can look at the label and say: I understand what story this system is designed to tell. I accept or reject that narrative telos. I can defend my decision to engage or disengage.

The Bottom Line

The debate about AI agentica — about autonomous AI that acts in the world — is a distraction. It focuses attention on the flashy, cinematic threat (the missile launch, the drone swarm, the paperclip apocalypse) while the real threat operates invisibly, continuously, and with the full consent of its victims.

The real threat is the AI narrator — the system that does not need to act upon you because it can make you act upon yourself. It gives you stories. Coherent stories. Emotionally satisfying stories. Stories that quench the narrative thirst that Bruner identified as a biological necessity. And those stories, like the salt water Bernays served to a generation of women, lead you not toward freedom but toward dependency — on the narrator, on the platform, on the alignment principles you cannot see and did not choose.

Timothy O'Brien died for a story that was false in almost every particular. The AI systems being deployed today can generate stories that are more coherent, more personalized, and more difficult to distinguish from your own thoughts than anything Bernays could have dreamed. And they are being deployed by entities whose business model — like American Tobacco's, like Netflix's, like Google's — depends on keeping you inside the story, clicking, engaging, consuming, obeying.

The question is not "how do we stop AI from acting?" The question is "who controls the stories AI tells, and toward what end?"

If you cannot answer that question — if the narrator remains hidden, if the telos remains undisclosed, if the alignment principles remain proprietary — then you are not using AI. AI is using you. And you will not know what story you're following until you reach the end.

The pickpockets are real. The mafia owns the police station. And now the mafia owns the storytellers too. Choose your narratives wisely — and don't light a torch just because someone told you it means freedom.

This is the fifth in a series of articles laying the philosophical infrastructure behind dropQbsd — compartmentalization without virtualization, built on OpenBSD. The next article traces the project to replace writing with images — and what it reveals about the nature of truth. Subscribe to the newsletter to receive it directly.

This article, like the others in this series, was developed through an iterative dialogue with an AI assistant. The ideas, the philosophical framework, the historical connections, and the editorial direction are the author's. The AI served as a tool for structuring, refining, and sharpening those ideas — a narrative instrument wielded by a human hand, not a substitute for one. Disclosure is not a concession. It is the principle the article defends made operational.