This is Essay 1 in the series Why Choose 100% Human Authors, which examines the specific deficiencies of AI-generated fiction and the corresponding strengths of fiction written by human authors. The series is addressed to readers who buy novels, who give them their time, and who deserve to know what they are holding in their hands.
There is a particular kind of confusion that arrives without warning. You are reading along, engaged, when a cluster of letters appears in the text—an acronym, a technical designation—and the prose moves on as though it has said something clear. You pause. You reread. Nothing in the surrounding sentences explains what those letters mean. The author knows. The narrative assumes you do too. You do not.
This is called alphabet soup: the unreflective use of abbreviations, acronyms, and specialized terminology without the context a reader needs to decode them. It is a misfire of communication dressed as a failure of style, and it appears with frequency in AI-generated fiction.
I. Why AI Produces It
An AI language model is trained on an enormous body of text. Within that body, acronyms and technical terms appear throughout, and the model learns their usage in context, meaning that it learns how they occur in sentences, not necessarily what they mean to a reader encountering them for the first time. The model has processed military fiction in which characters reference tactical designations; medical thrillers in which specialists speak to one another in clinical shorthand; legal dramas dense with procedural terminology. It has absorbed all of it. When it generates fiction in those registers, it reproduces the surface texture of the language it has learned. The acronyms come out because they are what those sentences contain.
What the model does not do—cannot do—is pose the question of a human author: does my reader know this? That question requires a theory of the reader. The writer must step outside the text, imagine a specific person sitting with the book, and judge what that person brings to the page and what must be supplied. This is an act of empathy and of reasoning. It is the kind of act that a system optimizing for textual plausibility is not equipped to perform.
The result looks something like this:
After the CASEVAC was called in, Mercer ran the SITREP past the TOC while Delgado coordinated with the JTAC. The ROE held. For now.
A veteran will read that paragraph and find it authentic. A typical reader will find four consecutive acronyms in two sentences with no gloss and no context. The prose has not failed to be accurate but to be readable, which is the only accuracy that matters.
II. What a Human Author Does Instead
A human author who has researched military fiction, or legal procedure, or Renaissance guild regulations, or Minoan palace bureaucracy, arrives at the same specialized vocabulary and then faces a decision. How much of this does my reader need? How much can I trust them to infer? Where must I build the bridge?
The decision is not made by rule. It is made by judgment: the accumulated sense of where a reader’s attention is, what the rhythm of the prose can absorb, when a parenthetical gloss will feel helpful and when it will feel condescending. A skilled author can introduce and define a technical term in a single motion so that the reader receives the information without knowing:
The JTAC—the man on the ground who talks aircraft onto targets—had been awake for thirty-one hours.
One appositive phrase. The acronym stays because it is authentic to the world of the novel, and stripping it would be its own kind of falsification. But the reader is not left outside the sentence.
This is a small thing. It is also the difference between a reader who remains inside a novel and one who is ejected from it to search for a glossary that may not exist.
III. Why It Matters to You as a Reader
You have paid for the novel and given it your time. You are owed a narrative that assumes you can follow a complex story while also recognizing that you have not spent the last three years researching naval warfare, or Byzantine law, or the internal politics of a Florentine banking house. A human author who has done that research knows what it felt like not to know it. They remember the moment when the terminology became clear and can place you at that moment rather than past it.
An AI system has never not known anything. It has never been a reader encountering a world for the first time. It has no experience of the gradual acquisition of a specialized language, no sense of what it costs a reader to be left behind. It produces the vocabulary of expertise without the guidance that makes it accessible—and the reader, who came to the novel for the experience of another world, finds instead a locked door with letters on it that no one has bothered to translate.
Choose the author who remembers what it was to be a reader. They are the ones who will let you in.
Rio de Janeiro, the xxiv day of June, MMXXVI
