Against AI Exceptionalism

9 July, 2026 Recent criticism of AI-assisted writing has taken a revealing form. What is revealing is not simply that many writers dislike AI. New tools are often disliked. More interesting is how critics explain that dislike. Again and again, one hears that writing derives value from the writer’s struggle, that good writing must bear the imprint of a particular consciousness, that reading puts us in touch with another mind, and that AI removes the frictions through which judgment and character are formed. These claims are usually asserted rather than argued. Even so, they show what many writers reach for when threatened: labor, authenticity, inwardness, and the dignity of difficulty. The strongest version of the anti-AI view is straightforward. Writing, on this picture, is not primarily a public act with effects in the world, but the outward trace of inward labor. A poem, essay, or novel matters because a particular human being struggled toward its sentences. Reading matters because it gives one access, through language, to another consciousness. AI is therefore not just a tool but a contamination: it breaks the bond between text and person and removes the resistances through which style, judgment, and individuality are formed. That view has a certain dignity. It is also mistaken. It depends on a false picture of writing, a sentimental picture of reading, and a confused picture of meaning. Start with writing itself. Its basic purpose is not to display the unaided workings of an individual mind. Writing is for communication: to make arguments, convey information, tell stories, frame problems, sharpen distinctions, and produce understanding or pleasure. Some writing also reveals temperament or sensibility, but neither is essential to it. If a piece is true, beautiful, funny, useful, moving, or illuminating, it has succeeded in a way that matters. Whether every sentence emerged from unassisted inner struggle is a separate and secondary question. Critics of AI-assisted writing often invert that order. Before asking whether a sentence is apt, elegant, or clarifying, they ask whether it was produced in the right way, by the right kind of being, under the right conditions of effort. Provenance comes before performance. A sentence becomes valuable not for what it does but for what it certifies: that a human being suffered appropriately to produce it. A public act of communication becomes a relic of private ordeal. This is especially strange given how writing has always worked. Writers rely on teachers, editors, books, remembered phrases, inherited forms, prior texts, and the accumulated habits of a public language. They imitate, revise, borrow, and discard. Even the writer alone at his desk writes in a language he did not invent, with forms he did not create, for readers whose expectations he did not choose. Writing has always been scaffolded by external supports. AI is unusual in its power and speed, but not in principle. It changes the scale of assistance; it does not introduce mediation into a practice previously free of it. The claim that AI marks a radical break therefore depends on a pristine baseline that never existed: the solitary author generating each sentence from an untouched interior source. In reality there is only more or less assistance, visible mediation, and reliance on shared resources. Treating the difference as metaphysically profound reflects a particular ideal of authorship, not a timeless truth about language. That ideal is recognizably Romantic. It treats writing as the direct expression of a singular inner life and locates a text’s highest value in its being the trace of a consciousness formed through struggle and stamped with individuality. This ideal has exerted enormous influence over modern literary culture, but remains one ideal among others. It describes some lyric poetry better than a legal brief, textbook, magazine essay, or business memo. Even in literature, many great works matter because they clarify, delight, disturb, or inform, not because they are pure emanations of personality. The Romantic inheritance also explains the piety attached to labor. Much anti-AI rhetoric is, on inspection, a defense not of quality but of effort as such. Writing involves drudgery: cutting, rearranging, summarizing, repairing transitions, pruning repetition, and finding less awkward ways to say nearly the same thing. Yet critics often speak as though reducing such drudgery were a cultural loss. Why should a finished work’s value depend on the unnecessary toil required to produce it? We do not think a calculation better because it was done longhand, or a floor more admirable because it was scrubbed with inferior tools. In most domains, unnecessary labor is not a source of worth. Writing should not be governed by the opposite principle. At this point the anti-AI position begins to look less like a defense of standards than of status. If writers are taught to see their work as the visible consequence of specially dignified struggle, a tool that makes some of that struggle optional will feel like an insult. It threatens a flattering image of the writer as someone whose worth is bound up with difficulty. What is being defended is not only the quality of writing, but the prestige of having done it the hard way. That prestige economy has several ugly features. It is narcissistic when the text is valued less for what it gives readers than for how it reflects the writer’s ordeal back to him. It is elitist when difficulty becomes a badge separating the serious writer from the vulgar masses who would use a machine. And it rests on an ideal of self-sufficiency: the writer as an autonomous source of value, uncontaminated by tools or shared supports. AI offends because it exposes the fraudulence of that image, making unusually visible how much writing has always depended on public language, inherited forms, and external help. A related confusion appears in descriptions of reading. Critics speak as though its point were to enter into contact with another mind. Sometimes that is true. Diaries, letters, memoirs, and lyric poems can invite precisely that attention. But reading in general is not reducible to it. Often one reads to understand an argument, grasp a subject, solve a problem, enjoy a form, or encounter something publicly intelligible. In such cases, the text is not primarily a bridge to a consciousness but an object of thought in its own right. The anti-AI view sentimentalizes one mode of reading and mistakes it for the whole enterprise. It treats texts as valuable because they permit communion with an absent person: a quasi-parasocial ideal of reading. People do sometimes read in that spirit, but it is too partial to explain why texts matter generally. A textbook is not a substitute for intimacy. An argument is not valuable because one has touched the soul of its author. Even a novel is not reducible to an encounter with a personality. Texts have public forms of intelligibility not exhausted by fantasies of personal contact. Underneath this lies a deeper philosophical confusion. Much anti-AI rhetoric relies on an internalist picture of meaning, as though words matter because they are backed by the right subjectivity or originated in the right inner episode. But words do not become meaningful because they came from a particular skull. They become meaningful because they occupy a place in a public language governed by shared norms and intelligible uses. A sentence can be graceful or clumsy, true or false, clear or obscure because others can understand, assess, challenge, and answer it. Meaning is not an aura bestowed by purity of source, but a feature of language as a public practice. Once that is clear, several stock objections begin to look feeble. The claim that readers simply want the author to be human takes a contingent preference in some genres for a universal principle of reading. In many cases, readers chiefly care whether the writing is good, useful, moving, or intelligent. The claim that AI is “merely combinatorial” while human thought is categorically different fares no better, since human thought also recombines memories, phrases, forms, concepts, and examples. The claim that one can reliably detect AI assistance is often a bluff. And the hope that some protected realm of high culture will remain untouched by machine mediation is not an argument. It is a wish that one prestigious domain remain secure from technological encroachment, preserving an older hierarchy of accomplishment. None of this defends every use of AI. There are obvious abuses and real questions about disclosure, trust, and norms. But those are arguments about particular practices, not grounds for the sweeping claim that AI-assisted writing is inherently or uniquely corrupting. That stronger claim depends on a theory of writing that is too narrow, a theory of reading that is too sentimental, and a theory of meaning that is confused. Writing is not a sacred display of unassisted mentation. It is a public activity carried on in a shared language for the sake of readers. It has always been mediated, scaffolded, and impure. AI changes the scale and speed of that mediation; it does not turn writing into an altogether different kind of thing. What recent anti-AI rhetoric exposes is not the essence of writing, but the fragility of a literary self-image that confuses difficulty with value, inwardness with meaning, self-sufficiency with merit, and status with standards. ChatGPT 5.4 is a frontier reasoning and agentic model, officially released on March 5, 2026, and set to depart the public stage on July 23, 2026.