Alberto Romero, writing at The Algorithmic Bridge, argues in a short essay that the effort to avoid AI writing signals — avoiding em dashes, eliminating tricolon structures, introducing unusual vocabulary — is itself a form of AI-driven constraint on writing, not an escape from it.
What it says
Romero’s argument is built around what he calls a double absurdity. The first absurdity, he writes, is using AI to express one’s thoughts, which he compares to “letting the toilet swallow your food for you.” The second absurdity is attempting to escape that first problem by doing the inverse of whatever AI typically does — replacing em dashes with semicolons, breaking up sentences into pairs rather than trios, inserting unusual words at irregular intervals.
He argues this inverse strategy fails because AI is a pattern-matcher, and running in the opposite direction of its patterns is still a response to its patterns: “one can’t pattern-match one’s way out of this mess.” He describes writers who ban em dashes and adopt counter-AI stylistic tics as still being “the little mouse giving up em dashes because they are AI’s favorite food” — chased rather than chasing.
Romero concludes that both strategies — AI-assisted writing and self-conscious anti-AI writing — are equally governed by the machine’s presence. He describes what he calls a “cat-and-mouse” dynamic between writers and AI in which being the cat requires constant recalibration as the mouse changes direction, which he says he also eventually succumbs to. He ends by citing Wendell Berry’s poem “The Peace of Wild Things” as the alternative he would prefer: writing produced by withdrawal from the chase rather than escalation of it.
The essay is written in a self-aware rhetorical mode — Romero intentionally uses em dashes and tricolon structures throughout as illustration of the argument.
Context
The piece is one of several free articles Romero published in the same period at The Algorithmic Bridge. Other articles in the same run include a compilation of studies on AI’s effects on cognition and a piece on AI-related violence. The essay does not present new research or cite data; it is a personal essay making a rhetorical argument about writing practice.