Today's computer scientists play the same role as the oracles of the ancient world and the astrologers of the Middle Ages. Modern predictions not only advise on war, crop output, and marriages, but algorithms and statisticians also now determine whether we can get a loan, a job, an apartment, or an organ transplant. And when we cede ground to these predictions, we lose control of our own lives.
In this powerful, refreshing new look at the many ways prediction shapes our everyday lives, University of Oxford professor Carissa Véliz explains how putting too much stock in others' predictions makes us vulnerable to charlatans, con artists, dubious technology, and self-deception. Examining a wide range of subjects both personal and societal, including medicine, climate, technology, society, and others, Véliz uncovers a number of predictions about humans tend to be self-fulfilling; more data doesn't guarantee better outcomes; AI is more likely to increase risk than decrease it; and a free and robust society requires not more prediction, but better preparation.
Véliz argues in this incisive and bracingly original book that the main promise of prediction is not knowledge of the future, but rather power over others. Prophecy is an invitation to defy those orders and live life on our own terms.
The Authority of What Has Not Yet Happened Carissa Véliz’s “Prophecy” traces prediction from sacred theater to algorithmic power, revealing how forecasts can become scripts before lives have had time to answer. By Demetris Papadimitropoulos | April 20th, 2026
In the cold half-light of “Prophecy,” an oracle-machine watches over an unsigned page, while the door waits to decide whether a life may enter its own future.
A forecast is never only a forecast once someone else has to live under it. It may arrive in modest clothes: a probability, a score, a risk label, an institutional hunch with decimal places and a frictionless interface. But once it helps decide who gets the loan, the job, the apartment, the hospital bed, the prison sentence, or a feed’s attention, it has stopped behaving like weather and started behaving like a script.
Carissa Véliz’s “Prophecy: Prediction, Power, and the Fight for the Future, from Ancient Oracles to AI” is sharpest at that threshold: where a guess gains power over a life. Prediction may fail, of course. The deeper trouble begins when it changes incentives, access, and behavior until the forecast starts to look fated. The person denied the chance is then told the system merely saw what was coming. How tidy. How ancient.
The book returns to that threshold until repetition becomes a form of pressure. Véliz has not written another solemn sermon about machines. She reaches backward into temples, omens, entrails, court astrologers, monarchs, con men, and the old commerce of dread. Her claim is that the prediction trade is not as modern as its vendors and evangelists prefer to imagine. Algorithms may run on data rather than laurel smoke, but they often perform an ancient bargain: fear sold back as authority. A prophecy, in this account, is not only a guess about what comes next. It is a maneuver of power with better lighting.
“Prophecy” begins by letting the reader hear the machinery under the floorboards. Perhaps an algorithm helped bring this book to your attention; perhaps a bookseller, librarian, friend, publisher, or publicist made a quieter forecast about your taste. The joke opens into a trapdoor. Predictions now shape credit files, job offers, housing decisions, medical care, online visibility, state force, and institutional punishment. Véliz is not arguing that all forecasting is foolish. Weather apps, epidemiologists, and doctors all have good reasons to look ahead. Her point is severe in the way useful arguments often are: when predictions about people acquire the tone that makes objection seem rude, institutions can punish us for futures that have not happened and may never happen. The question is not only whether the model is accurate. It is who has been allowed to turn a guess into a gate.
Before the closed door, the figure has not yet acted, but the score has already learned how to stand in for judgment.
First “Prophecy” seduces, then it indicts, then it asks how to live without kneeling. Oracles, astrologers, probability, measurement, risk, and AI as the contemporary Oracle of Delphi establish prediction’s glamour. Forecasts mistaken for facts, guesses hardened into verdicts, surveillance systems that turn people into profiles easier to price, rank, and police, and the stubborn survival of the unforeseeable expose its damage. The final movement turns toward truth, virtue, beauty, creativity, humor, resilience, democracy, Epicurus, and the art of not outsourcing one’s courage to a forecast. The sequence teaches the reader how to distrust without merely curling into paranoia. It is not a retreat from foresight. It is a refusal to confuse foresight with rule.
Véliz is especially good on the old theater of prophecy: the strange, lucrative, occasionally ridiculous business of knowing before others do. Tiberius tests astrologers and throws failures off cliffs. Delphi becomes sacred site, tourist economy, political theater, meat market, and prestige machine. Alexander of Abonoteichus manufactures the snake-god Glycon through masks, tubes, sealed-scroll tricks, charisma, and stagecraft good enough to pass for revelation.
These episodes are funny because human beings are funny when frightened, and chilling because frightened people are profitable. The old material is not decorative. It teaches the reader where to look first. Follow the fear. Follow the money. Follow the person who claims the world has already been read.
Prophecy begins as theater and invoice: a false god, a sealed scroll, and the old human wish to buy certainty from the dark.
That long backward glance earns its keep. Véliz makes us sit with older prophets before she lets the new ones enter the room. By the time the tech executive appears as court oracle, the comparison has earned its robe. Artificial intelligence does not look like a clean rupture from the past. It looks like the latest costume worn by the same triangle of dread, money, and power, plus the promise that someone else can tell us what we are about to become. The robe fits for a long while. When the cloth pulls, the strain matters.
The prose keeps scraping the polish off prediction. Véliz writes in sharp declarative bursts when she wants the argument to strike: predictions are not facts; prediction is power; the future is not yet available for inspection. Elsewhere, she opens into longer, theatrical passages full of temples, smoke, Oxford dinners, collapsing ceilings, fraudulent snakes, terrified rulers, and modern executives enjoying the ancient pleasure of being consulted. Her diction is clear enough to travel fast, charged enough to cut. She likes words with moral weight: truth, power, virtue, beauty, verdict, command, script. She also has a comic instinct that keeps the book from becoming sermon-shaped. The jokes do not land with equal delicacy, but the liveliness matters. A neutral version of this argument would be tidier and far less memorable. This one has elbows.
The style carries part of the case. Véliz is attacking neutrality with a thumb on the scale. Her own voice does not pretend to be a voice without fingerprints. It is learned, impatient, witty, occasionally severe, and willing to let a grotesque image do honest work. That choice gives the book personality, but it also gives it a method. “Prophecy” wants to de-sanitize prediction. It wants the model to smell faintly of incense, money, sweat, and institutional carpet.
Véliz’s sharpest move is to ask not whether a prediction works, but who has permission to let it decide. Many books about algorithms stop at accuracy: did the system classify correctly, predict reliably, malfunction visibly? Véliz asks who gets to predict whom, with what data, under what authority, for whose benefit, and at whose cost. That is more disturbing than the familiar story in which a model reads the past badly and someone loses a chance. A risk score may be statistically defensible and still morally suspect. A forecast may be plausible and still abusive. A system may sort quickly and call it wisdom.
Here “Prophecy” belongs near “Weapons of Math Destruction” by Cathy O’Neil, though Véliz is less case-based and more philosophical. It glances toward “The Black Swan” by Nassim Nicholas Taleb in its impatience with predictive arrogance, though her concern is less market uncertainty than power over other people’s chances. In its movement from technological critique to the possibility of living otherwise, it has a distant kinship with “How to Do Nothing” by Jenny Odell, though Véliz is more combative, less meditative, and more likely to bring an unsafe emperor to the party. Useful, briefly. Then someone should hide the knives.
The strongest chapter may be “When Predictions Become Verdicts.” Véliz begins with Caracalla and Macrinus: a prophecy that Macrinus will become emperor helps create the conditions under which Macrinus has Caracalla murdered and succeeds him. It is the book’s thesis in miniature. The prediction does not simply foresee the future; it intervenes in it. From there, Véliz develops self-fulfilling prophecy across education, business, finance, racism, sexism, venture capital, publishing, and surveillance. Her most forceful claim is that predictions about human beings can operate as statements that rearrange behavior. They appear to describe, but they also instruct, pressure, rank, invite, exclude, and command. A forecast can become a leash while still calling itself a lens.
An empty chair, a strip of paper, and a tightening shadow suggest the moment a forecast stops describing the future and begins arranging it.
The problem does not need to be imported from the headlines; the headlines merely make visible what the book already knows. We live among systems that rank us before they hear us. People are priced, filtered, recommended, ignored, targeted, nudged, deprioritized, and sometimes punished by predictions they cannot see, question, or correct. Véliz’s point is not that every model is predatory, or that every attempt to anticipate risk is tyranny in a white coat. Her better point is that prediction grows most dangerous in the places where paperwork already resembles care: hospitals, hiring systems, housing decisions, insurance, policing, feeds, institutional triage. The forecast says it is helping. The person being forecast into a corner may experience it as a door closing from the other side.
In the institutional green of “Prophecy,” people wait beneath quiet lights while unseen systems price, rank, and explain them before they speak.
The analogy eventually shows its stitching. A weather forecast, a medical prognosis, an actuarial table, a hiring algorithm, a credit score, a venture capitalist’s hunch, a chatbot’s probabilistic answer, and an oracle’s riddle all belong to the same unsettled province. They are not the same citizen. Their methods, safeguards, uses, and harms differ. Véliz knows this, and she acknowledges that some forecasts are justified and useful. But the rhetorical current runs hottest when prediction is being prosecuted as power. That heat clarifies. It also sometimes singes distinction.
The same reach that gives the book its sweep also lets too many anxieties through the door. Véliz openly wants “Prophecy” to be hard to classify: technology criticism, business book, political treatise, philosophy, partial memoir, academic critique, literary defense, humor-inflected public argument, and guide to living well. That reach gives the book range, but not every strand receives the same oxygen. The ancient material, the AI-as-oracle argument, the sections on self-fulfilling prophecy, and the closing defense of uncertainty feel essential. Some detours, especially when the target field widens to effective altruism, academia, publishing, climate, comedy, and every anxiety about what comes next, can feel less like escalation than annexation. The book never coasts, but it sometimes arrives carrying more luggage than the room can gracefully hold.
The repetition is real, though not careless. The claim that prediction is power, not knowledge, is excellent. It deserves to travel. It does travel. Then it circles the block, waves from the corner, and appears at the kitchen door with a casserole. Véliz is crossing so many domains that the thesis has to reappear in new clothes. Still, a reader may occasionally reach the destination before the paragraph does.
Yet the book would be poorer without its appetite. “Prophecy” works because Véliz understands prediction as more than a technical problem. It is a problem of fear, profit, and permission. We want forecasts because we are fragile, desiring creatures who know the last fact of the story but not its timing, texture, or terms. We want warnings. We want reassurance. We want to know whether our children will be safe, whether our work will last, whether our bodies will betray us, whether our countries will hold, whether the systems humming around us are sorting us toward shelter or ruin. Into that vulnerability comes the prophet, ancient or modern, with a fee schedule and a confident tone.
By the end, warning gives way to a harder question: how much of the future must remain unassigned? Suspicion is not the destination, only the tunnel out of false obedience. A weaker version of this book would end by scolding AI and waving a little flag marked ethics. Véliz instead asks how to live when the future cannot and should not be mastered. The final chapters argue for creativity, humor, resilience, democratic openness, and Epicurean attention to the life not yet converted into forecast. Uncertainty becomes not the embarrassing remainder left after insufficient data, but the condition of freedom. The book’s closing image of the good life as a blank page rather than a script retroactively sharpens the whole project. Prediction is dangerous not only because it may fail, but because it may steal the future by pretending to reveal it.
After the machinery of prediction recedes, the blank page remains: not proof of safety, but the fragile mercy of an unwritten life.
That turn keeps “Prophecy” from closing as another dispatch from the kingdom of smart machines and foolish humans. It does not merely ask whether machines can know us. It asks what is lost when being known in advance becomes the toll for admission. Why has the future become a credit file, a medical queue, a recommendation feed, a risk category, a moral calculation, a hiring screen, a market signal, an assigned role? And what would it mean to refuse a role prewritten by someone else’s model?
My final rating is 86/100, which corresponds to 4/5 stars. That feels like the right temperature: admiring without genuflection; persuaded by the governing lens, aware that the analogy stretches; grateful for the wit and sweep, unconvinced that every strand needed equal stage time. “Prophecy” is strongest when forecast meets gatekeeping power. It is less strong when all prediction is made to stand too near domination. But even its overreach is the overreach of a book trying to think at scale, not slackness.
What remains after the arguments, emperors, algorithms, fraudulent snakes, tech prophets, and philosophical counterspells is a useful irritation toward being sorted. The next time a system predicts you, recommends you, rejects you, scores you, ranks you, or tells you what people like you are likely to do, Véliz’s book may whisper its most useful warning: do not mistake the forecast for the life. Others may model, monetize, and market the future. They do not get the final say merely because they built the model. The middle of the page is still open, and no oracle has yet signed its name.
Early thumbnail studies test the room’s argument in miniature: desk, door, orb, page, and shadow arranged until prediction begins to feel like architecture.
The first pencil structure fixes the room’s quiet geometry before color enters: the door as gate, the page as future, the orb as watchful doubt.
The cover palette is tested as atmosphere: acid chartreuse, celadon, teal, aqua, and black-green held in tension before they become the room’s uneasy weather.
The first wash lets the room begin to breathe, softening the pencil’s administrative geometry into the cool, suspended dread of “Prophecy.”
Small motif studies refine the image’s symbolic grammar: the orb watches, the door withholds, and the blank page waits without surrendering its silence.
Instead of a visible character, this shadow study tests the human presence as absence: someone forecast, delayed, and almost but not quite erased from the room.
All watercolor illustrations by Demetris Papadimitropoulos.
This is a wonderful book about the ethics of prediction - of how predictions shape the world we live in and the implications of embracing prediction machines (AI) in many spheres of life. The book is very well grounded in philosophy and some history, yet not heavy - it’s the kind that can be recommended to any intellectually curious mind as long as they are fond of reading. If you’re a researcher, you might find the grounding useful as it can serve as a foundation or scaffold for your own thinking. One of the reasons for that it because the arguments stand independent of the type of AI technology in question. This is a point that many techo-optimists miss- technology is not a panacea. If it were, no child would starve (we have the technology and capacity, and, in fact, produce enough food to feed everyone. The problem is social). In any case, it’s a thoroughly enjoyable read and I strongly recommend it.
Truly remarkable book, written very personally and honestly. Véliz explores prophecy throughout the ages, drawing towards the similarities between ancient seers and modern tech leaders.
This was pretty thought provoking. I wish it hadn’t so explicitly dealt with AI in a few places because it’s already outdated and the argument feels weaker for it, even if it’s actually not.
A brilliant recommendation by my Philosophy & Ethics lecturer at Uni of Birmingham. An interesting counter to William Macaskill who I equally enjoyed reading.