CONCEPT
The Active Voice Discipline
Strunk’s grammatical rule transmuted into a practice of visibility in the AI age—the insistence on naming who did what, which prevents human agency from disappearing into the passive constructions that large language models produce by default.
The active voice, in William Strunk Jr.’s original formulation, is “usually more direct and vigorous than the passive,” and his examples are grammatical: “I shall always remember my first visit to Boston” is better than “My first visit to Boston will always be remembered by me.” The rule is a stylistic one, but its deeper content is ethical: name the agent, assign the action, say who did what. “Mistakes were made” is the most famous passive construction in American political language because it admits the existence of mistakes while concealing who made them—a grammatical evasion that the active voice would prevent.
Large language models produce passive constructions systematically and at scale, because they are trained on institutional text—academic papers, corporate communications, government documents—in which the passive is the default precisely because institutions prefer not to name agents. When a writer accepts AI-generated prose without applying the active-voice filter, she accepts, along with the prose, its embedded institutional evasion: