The term aiphobia is a compact neologism for a fear already circulating under names such as AI phobia. It is built the way aiciety builds its words — by fusing artificial intelligence with something older and more human, in this case the Greek phóbos, fear. It names the unease that rises in a person the moment a machine begins to do something we had reserved for ourselves: to write, to compose, to diagnose, to judge. But unlike many coinages, aiphobia describes nothing entirely new. It is the latest verse of a very old song, and the song has always been less about the instrument than about the people listening to it.
The oldest new fear
Around 370 BC, in a dialogue Plato set on the banks of the Ilissus, the god Theuth presents the Egyptian king Thamus with a gift: writing, which he offers as a remedy for forgetfulness and a sharpener of wisdom. The king declines to be grateful. Writing, he warns, will not strengthen memory but hollow it out. People will trust external marks instead of their own minds, and they will acquire the appearance of knowledge without its substance — reading much, understanding little.
It is worth sitting with how completely wrong, and how partly right, Thamus turned out to be. Writing did not destroy human thought; it became its scaffolding. Yet his deeper intuition — that a cognitive technology reshapes the mind that uses it, and not always in the direction of wisdom — was not foolish. It is one of the earliest canonical records of this pattern: not aiphobia yet, of course, but the same anxiety aimed at writing itself.
The pattern repeats with almost tiresome reliability. The printing press would spread heresy and corrupt morals. Railway travel would unsettle the nerves and derange the body at speed. The telephone would invite strangers into the home; the calculator would weaken arithmetic; the computer would abolish work. Each fear was eventually absorbed, and each was never entirely groundless — it simply aimed at the wrong target. That is the signature of phobia as opposed to reasoned concern: not that the worry is baseless, but that it fastens onto the visible object instead of the real one.
Phobia is not the same as concern
Here a distinction has to be drawn, because the public conversation collapses it constantly. To ask whether an opaque model is denying people loans along lines that track race is not phobia; it is diligence. To ask who owns the most capable systems, and what that concentration does to power, is not phobia; it is political literacy. Reasoned concern scales its response to the evidence. Phobia does the opposite: it takes a genuine and bounded risk and inflates it into a totalizing dread, or it imports a dread from fiction and dresses it as analysis.
The clearest tell is the Skynet reflex — the assumption that a system trained to predict the next token harbours a will, a survival instinct, an appetite for our subjugation. There is no good evidence that today’s mainstream language models possess any of these. They are extraordinary engines of pattern, not minds with desires. Mistaking the one for the other is not merely a category error; it is expensive. Every hour spent fearing a machine uprising is an hour not spent on the unglamorous harms already at the door: the manipulated voice note, the silently biased ranking, the slow outsourcing of one’s own judgement to a confident interface. The near-term danger of AI lies less in its malice than in our complacency.
What the fear is actually about
If aiphobia were really about the machine, it would have a different shape. We do not panic at a chess engine that beats every grandmaster alive, nor at a calculator that humiliates us at long division. We tolerate, even celebrate, machines that exceed us at narrow tasks. The fear sharpens only when the machine encroaches on the territory we had used to define ourselves: language, creativity, judgement, the sense of an inner life.
This is why the Japanese roboticist Masahiro Mori’s “uncanny valley” is the right metaphor for the wrong reason. The eeriness of the near-human is not really about the object’s imperfection. It is about the mirror it holds up. A system that writes a passable condolence note, or paints in your late grandmother’s style, may threaten more than livelihood. It threatens a premise — the quiet assumption that these were the things only a soul could do. Aiphobia, at its root, is less a fear of artificial intelligence than a fear of human ordinariness; less a horror of what the machine is than a vertigo about what we are not.
The Luddites are remembered as the patron saints of this fear, and they are remembered wrongly. The weavers who broke the frames in the 1810s were not confused about how machines worked, nor opposed to technology as such. They were protesting a specific arrangement: machinery deployed to break skilled labour on terms its owners alone had set. Their quarrel was with the distribution of power, not with the gears. We flatter ourselves when we file every objection to AI under irrational technophobia. Some of it is the Luddite’s actual question, still unanswered: not can this be built, but who decides on whose terms.
The danger is not technological
Heidegger, writing about an earlier wave of machinery, made an observation that aiphobia keeps proving true: the essence of technology is nothing technological. The danger he located was not in any device but in a way of seeing — a habit of mind that converts everything, eventually including people, into raw material to be optimized and stored. Read in this light, the truly worrying thing about a society saturated with AI is not that the systems might rebel. It is that we might quietly adopt their logic: treating attention as a resource to be extracted, judgement as a function to be automated, the friction of thinking as a bug to be patched.
That is a real risk, and it is precisely the one aiphobia is poorly equipped to see, because it is looking the wrong way — outward, at the artifact, when the relevant transformation is happening inward, in us. A fear trained on the robot uprising will not notice the slow erosion of the capacity to sit with a hard question unaided. The call, as ever, is coming from inside the house.
Neither fear nor worship
Spinoza observed that hope and fear are not opposites but twins — there is no fear that does not carry a shadow of hope, and no hope unhaunted by fear. The breathless AI enthusiast and the trembling AI catastrophist are closer than either would admit: both have surrendered the same thing, which is agency. Both treat the arrival of aiciety as weather — something that happens to us, to be either welcomed or dreaded, but not shaped.
The way out of aiphobia is not optimism, which is merely fear’s more cheerful sibling. It is understanding, in the unglamorous, literacy-shaped sense. Whoever grasps that these systems run on probability rather than magic, on statistics rather than secret intentions, has less to fear and a sharper eye for what genuinely warrants fear. Understanding dissolves the phobia and preserves the concern — which is exactly the right trade.
This is the wager beneath the whole idea of aiciety: that the society now forming is not arriving on its own. It is being assembled out of countless human decisions — in legislatures and boardrooms, in classrooms and on the contact surface of a single prompt. Aiphobia matters not because the fear is shameful, but because an unexamined fear makes those decisions for us, in the dark. To understand the fear is to take the decisions back into the light.
Perhaps, then, the best answer to aiphobia is not less feeling but more authorship. The machines are not coming for us. We are writing them — and, as Thamus knew, whatever we write, we are also rewriting ourselves.
Leave a Reply