Why Does AI Prose Forget That People Have Bodies?
This is Essay 17 of 22 in the series Why Choose 100% Human Authors, addressed to readers who want to understand what a human author does that a machine cannot and why that difference matters to the experience of literary fiction.
The Sensory Void
There is a kind of scene that recurs throughout AI-generated fiction: two or more characters speak to one another, decisions are reached, emotions are stated, and the scene concludes, having occurred nowhere in particular and cost nobody anything physical. The characters have not felt the temperature of the room. They have not noticed the ache of standing too long. They have voices and intentions—not bodies.
This is the sensory void: a scene constructed from speech and decision, in which the fact that a person exists in space and time has been omitted. The omission is seldom total. A character may be described once with a fixed inventory of features and never again does the body reassert itself. It is furniture established and then forgotten.
What an Embodied Scene Requires
Consider an invented example: two siblings meet to discuss the disposition of their late father’s house. In the version without a body, they speak in turn. One argues for selling; the other resists. Positions are stated; a compromise is reached.
Now introduce the body. The meeting takes place in late afternoon, and the sister who wants to sell has been on her feet since dawn. Her lower back hurts in a way that keeps interrupting her own sentences. The brother has not eaten since morning and finds himself arguing partly because he is hungry. Neither says any of this aloud, but the reader who has been given these details reads the same disagreement as something happening to two tired, present people, rather than between two positions wearing their names.
The outcome of the scene need not change. What changes is whether the reader has been present for it or only informed.
Talking Heads
A related failure is the talking-heads dialogue: the voices are clear but the bodies producing them are not. The reader can follow who said what but not see where anyone is standing, what they are doing with their hands, and whether they have moved closer or further away in the course of the exchange. The conversation floats, disconnected from any physical space.
An invented example: three characters in a kitchen, in conversation about whether to confront a fourth family member. In the talking-heads version, each line of dialogue is attributed and nothing else is given. The reader knows the content of the argument and nothing of its shape. Now add the missing bodies: one character has not stopped peeling the same vegetable for several minutes past necessity. Another has taken a position nearer the door than the conversation requires. The third takes a long time to answer a direct question because they have just become aware they are gripping the edge of the counter. None of this is stated as meaning. It is what the bodies are doing while the voices argue, and the reader assembles the meaning without being told to.
Human Writers Are Not Immune
It would be dishonest to present this as a failure exclusive to machines. Human writers produce sensory voids and talking heads as well, in early drafts or dialogue-heavy scenes where the temptation is to let the exchange carry on with wit or tension alone. The first draft of an argument scene is often two voices in a white room.
The distinction is not that human authors avoid this error but that they can recognize and correct it. They have a body and know what it does during an argument. For them, revision involves returning to a scene that works and asking what the room felt like, what the characters’ bodies were doing while their voices did the work the draft accomplished.
A language model has no such standard to measure against. It has read a great many scenes in which other writers solved this problem, and it can reproduce the solution as a pattern: a hand placed here, a temperature mentioned there. But it cannot feel the absence the way a writer feels it. It has never felt anything.
What a Human Author Does
A writer brings that knowledge to every scene without needing to be reminded of it. The character who has been arguing for an hour grows tired in a way the writer recalls from personal experience. The character standing in the cold shifts their weight as the writer has stood there and knows what the body does when it is asked to wait. This does not need to be deployed in every sentence. It surfaces because the writer cannot help but know it, and the prose carries its traces whether or not the writer thinks about it.
This is not a technique that can be applied as a rule. It is the residue of having a body and having paid attention to what that body does under the pressures the story asks the characters to bear. The reader perceives the residue without being able to say why. They have not been told the scene is physical. They have been present for it through a body that recognizes what another body felt.
Rio de Janeiro, the i day of July, MMXXVI.
