Speech genres are, in Bakhtin's framework, the shared forms that organize communication within specific domains of human activity. Every profession, every social context, every recurring situation develops its own genre: characteristic ways of beginning and ending, typical lengths, expected tones, conventional structures. The business email is a speech genre; so is the academic lecture, the medical consultation, the first date. Learning to operate in a domain means internalizing its speech genres — not just vocabulary but the social logic of what can be said, what goes without saying, what must be said with care. Speech genres are normative (they establish what counts as competent performance) and creative (they provide the constraints against which genuine originality becomes possible). Bakhtin distinguished primary genres (everyday, oral, dialogic) from secondary genres (written, formalized, often monologic) and argued that the novel's achievement is its incorporation of primary genres into secondary form, maintaining the living responsiveness of speech within the architectonic structure of literature.
Bakhtin introduced speech genres in his 1952–1953 essay (published posthumously in 1979, English translation 1986). The concept was partly a response to Saussurean linguistics, which treated language as an abstract system without accounting for the conventional patterns of actual use. Bakhtin argued that competence in a language means not only knowing grammar and vocabulary but also knowing which speech genres are appropriate to which situations — a form of social knowledge that grammars do not teach. Children acquire this implicitly; professionals develop specialized genres through training and practice; failures of communication often trace to genre mismatch (using a formal genre in an informal context, or vice versa).
The AI prompt is an emerging speech genre. It has developing conventions: be specific but not over-constrained, provide context but don't waste tokens, iterate rather than seeking perfection on the first try. Online communities (Reddit, Twitter, Discord) are collectively discovering these conventions through trial and error, and the genre is stabilizing even as the tools evolve. The machine's response is also genre-governed: it tends toward the expository, the pedagogical, the comprehensive — genres that dominate its training data. The builder must learn to recognize when the machine's default genre is inappropriate for the task and adjust accordingly (requesting a different tone, specifying a target audience, providing examples of the desired register).
The Orange Pill exhibits genre-awareness when Segal describes rejecting Claude's outputs that were too smooth, too literary, too structurally perfect. The machine's default is a secondary genre: polished, finished, suitable for publication. Segal's voice required a hybrid: the accessibility of primary genres (direct address, personal anecdote, roughness) combined with the architectonic ambition of secondary ones (sustained argument, cumulative structure, theoretical precision). Navigating this hybrid required constant adjustment — accepting the machine's structural contributions while rewriting for voice, keeping its connections while roughening its prose, maintaining the dialogue's discoveries while ensuring the text sounded like a person rather than a committee.
The institutional challenge is that organizations and educational systems do not yet teach genre-awareness for AI collaboration. Most training focuses on technical proficiency (how to write effective prompts) without addressing the rhetorical and social dimensions (which genres the machine defaults to, which it handles poorly, how to specify genre explicitly when the default is wrong). The result is competent but generically narrow AI use: everyone gets expository prose in the machine's preferred register, whether or not that register serves the communicative purpose. The alternative is to develop genre literacy for the AI age — the capacity to recognize, name, request, and evaluate the speech genres that structure human-AI dialogue.
Bakhtin wrote 'The Problem of Speech Genres' late in his life, during his partial rehabilitation in the 1950s. The essay was first published in Russian in 1979 (two years after his death) and in English in Caryl Emerson and Michael Holquist's Speech Genres and Other Late Essays (1986). It has since become foundational to genre studies, rhetoric, and applied linguistics.
The concept's application to AI-mediated communication is a 2020s development, though the framework invites it: if every domain has characteristic speech genres, then human-AI collaboration will develop its own, and understanding those genres is essential for competent use.
Genres organize utterances. Relatively stable forms corresponding to recurring social situations.
Competence requires genre knowledge. Knowing a language means knowing which genres are appropriate when.
AI prompting is an emerging genre. Conventions are stabilizing through collective discovery.
The machine defaults to specific genres. Expository, pedagogical, comprehensive — reflecting its training data.
Genre literacy is a new essential skill. Recognizing and specifying genres appropriate to the task improves AI collaboration.