If You Meet an Author on the Road, Ki*l Him

Abstract

The twentieth century repeatedly attempted to dethrone the Author without succeeding in eliminating the cultural desire for authorship. Roland Barthes displaced the writer as the sovereign origin of textual meaning, while Michel Foucault demonstrated that the author persisted as a classificatory, institutional and juridical function even after the metaphysics of individual creation had been destabilised. The emergence of generative artificial intelligence transforms this theoretical dispute into a material problem. Large language models can now perform operations previously treated as visible evidence of authorship: generating sentences, restructuring arguments, imitating genres, proposing metaphors, editing style, producing titles and transforming existing texts. Yet linguistic production does not necessarily establish authorial sovereignty. This article uses the deliberately provocative injunction If You Meet an Author on the Road, Kill Him as a metaphor for the repeated dismantling of every figure elevated into the final source of textual authority. The formulation adapts the Zen logic of destroying attachment to the master rather than advocating the destruction of the person: what must be “killed” is the intellectual idol of the sovereign Author. From this perspective, Barthes’s Scriptor, Foucault’s author-function, the contemporary Prompter and the generative Model represent successive attempts to relocate rather than finally resolve the problem of textual agency.

The article argues that generative AI exposes a fragmentation that had always existed inside the apparently unified act of writing. Intention, research, memory, linguistic formulation, recombination, editing, selection, publication, interpretation and responsibility need not belong to a single agent, and contemporary human–AI production makes their separation increasingly visible. The resulting condition is described as authorship without a centre: a communicative work can emerge through interactions among human intention, inherited language, computational prediction, machine generation, editorial judgement, institutional framing and reader interpretation without any single participant containing the complete causal or semantic origin of the finished text. This does not mean that writers, intentions or responsibility disappear. On the contrary, the dissolution of sovereign authorship makes responsibility more important because distributed production cannot become an excuse for unaccountable publication. The article therefore distinguishes the death of the Author from the disappearance of the human writer and the fragmentation of authorship from the abolition of accountability. Its central claim is that the death of the Author should not be understood as a completed historical event but as a recurring critical operation. Whenever textual culture installs a new sovereign centre—the genius, the Scriptor, the Prompter, the algorithm or the Model—the same theoretical question returns: whether the production of language grants authority over its meaning. In the age of generative AI, the task is therefore not to discover who has inherited the Author’s throne, but to question why the text requires a throne at all.

Keywords

authorship; death of the author; death of the reader; Roland Barthes; Michel Foucault; author-function; Scriptor; Prompter; generative artificial intelligence; large language models; distributed authorship; authorship without a centre; textual authority; human–AI writing; interpretation; responsibility

Introduction: The Author Who Refused to Die

“An author’s name is not simply an element in a discourse; it performs a certain role with regard to narrative discourse.”

— Michel Foucault (1969), “What Is an Author?”

The Author has proved remarkably difficult to kill. More than half a century after Roland Barthes announced the theoretical conditions for his death, contemporary culture appears more fascinated than ever by the question of who stands behind a text. Literary biographies reconstruct the private lives of writers in search of interpretative keys; publishers transform names into brands; universities police attribution; copyright law assigns ownership; journalism attaches responsibility through the byline; readers ask whether a novel expresses the convictions of the person whose name appears on its cover. Generative artificial intelligence has intensified this attachment rather than dissolving it. The first question asked about an unfamiliar piece of prose is increasingly not what it means but whether “a human wrote it”. AI-detection services, declarations of machine assistance, disputes about academic integrity and controversies surrounding generative literature all reproduce an ancient desire to locate language inside an identifiable origin. The irony is striking. At precisely the historical moment when textual production can be technologically detached from the individual hand, culture becomes increasingly anxious to recover the hand behind the text. This article therefore distinguishes from the outset between the writer and the Author. The writer is an empirical person who may research, remember, imagine, compose, revise, select and publish. The Author, capitalised here as a theoretical category, is something more ambitious: the imagined sovereign centre from which the text supposedly receives its unity, originality, intention and ultimately its meaning. To challenge this Author is not to deny that human beings write books or that their biographies can matter. It is to challenge the assumption that textual meaning must finally return to a single consciousness that precedes and governs the work. The provocative title If You Meet an Author on the Road, Kill Him converts this challenge into a methodological principle. Its violence is entirely metaphorical. It borrows the paradoxical logic of the Zen injunction to destroy attachment to the master encountered as an external authority: the obstacle is not the living person but the idol constructed around that person. Applied to writing, the command becomes an invitation to recognise the Author whenever criticism, technology or cultural institutions reconstruct him as the final explanation of language—and then to dismantle that sovereignty again.

The historical trajectory considered here consequently does not proceed towards a new claimant to the abandoned throne. It moves instead through successive failures to keep the throne empty. Barthes displaced the Author by describing writing as a field assembled from already existing cultural codes rather than the transparent expression of an originating consciousness. His Scriptor seemed to replace creation with inscription and originality with recombination, while simultaneously transferring interpretative possibility towards the reader. Foucault immediately complicated the funeral by asking what cultural work continued to be performed by the author’s name. The resulting author-function demonstrated that the disappearance of sovereign intention did not eliminate classification, attribution, ownership, prohibition, prestige or responsibility. The Author could die philosophically while remaining institutionally indispensable. The digital environment complicated the relationship further by fragmenting reading, circulation and textual stability, while generative AI now attacks what seemed to be the Author’s final material refuge: the act of linguistic production itself. A Prompter can initiate a work without composing most of its sentences; a Model can formulate language without independently deciding why the work should exist; an editor can transform machine output; an institution can publish it; a reader can produce meanings anticipated by neither human nor Model. The sequence Author → Scriptor → author-function → Prompter → Delegated Author therefore should not be mistaken for an evolutionary succession in which each new figure becomes the legitimate sovereign of the text. Each is instead evidence that what culture calls authorship contains operations that can be separated and redistributed. The central problem of generative AI is consequently not whether the machine deserves to be crowned the new Author. Such a coronation would merely reproduce the metaphysics that twentieth-century criticism attempted to dismantle. The more consequential possibility is that textual production is entering a condition in which no single participant can plausibly contain the complete origin of the work. This article calls that condition authorship without a centre. Its argument moves from the destruction of the authoritative master, through the theoretical death and institutional resurrection of the Author, towards AI-generated language in which intention, formulation, selection, interpretation and responsibility can occupy different locations. The question is no longer simply who wrote the text. It is why, after repeatedly discovering that language exceeds every individual who produces it, culture continues searching for someone to place above it.

Part I: Kill the Master — Zen, Authority and the Dangerous Author

“If you meet the Buddha, kill the Buddha; if you meet the patriarch, kill the patriarch.”

— Linji Yixuan, The Record of Linji

The violence of the Zen injunction is philosophical rather than physical: what must be destroyed is the attachment that transforms a teacher, doctrine or representation into an external authority standing between the practitioner and direct understanding. The master becomes dangerous at precisely the moment when reverence for the master replaces the experience towards which the teaching was intended to point. Read in this sense, the command attributed to Linji provides an unexpectedly precise entry into the history of modern authorship. Western literary culture gradually constructed around the empirical writer a comparable figure of authority: not merely the person who happened to compose a text, but the privileged consciousness imagined to exist behind it. The distinction is fundamental. A writer can possess intentions, memories, political commitments, private experiences and recognisable stylistic habits without thereby becoming the sovereign explanation of everything the work can mean. The Author emerges when these biographical realities are elevated into a hierarchy in which the text is treated as secondary evidence of an originating self. Reading then becomes a journey backwards. The poem points towards the poet; the novel becomes evidence of the novelist; ambiguity becomes a puzzle whose solution supposedly exists somewhere in biography, correspondence, interviews, notebooks or declared intention. The name printed on the cover begins to function like the name of the master. One asks what Shakespeare believed, what Dostoevsky intended, what Kafka “really meant”, as though meaning were a possession deposited by an individual consciousness into language and recoverable through sufficient proximity to its owner. Modern publishing intensified this structure by attaching the name to ownership, copyright, reputation, marketing and cultural prestige. Romanticism supplied the mythology of genius; the literary marketplace supplied the signature; criticism supplied biography; copyright supplied legal possession. These developments did not invent writers, but they helped consolidate the culturally specific figure of the Author as simultaneously origin, proprietor and interpretative authority. Journalism created an especially revealing version of the same structure through the byline. Here the authorial name performs a legitimate ethical function because statements require accountability, yet accountability can easily be confused with semantic sovereignty. The journalist must answer for what is published under a name, but this does not mean that the journalist controls every meaning subsequently produced by the text. The difference between responsibility and sovereignty is therefore already present before AI appears. To “kill” the Author in the Linjian sense is not to erase the writer’s labour, deny intellectual ownership or abolish responsibility. It is to break the attachment through which the empirical producer of language becomes its final master.

The mythology of sovereign authorship nevertheless remained persuasive because it offered an intuitively satisfying model of creation. A work seemed to travel from inside to outside: an idea originated in the private interiority of an individual, language gave that idea material form, and the completed text carried something of its creator into the world. The model resembles a theological account of creation because the creator logically precedes the created object and therefore appears to possess privileged knowledge of its purpose. The vocabulary surrounding artistic production still preserves traces of this structure: the author “conceives” the work, gives “birth” to it, possesses a distinctive “voice”, and leaves a personal “imprint” upon the material. Yet writing itself continually resists this mythology. No writer invents the language in which the work is produced. Genres precede individual works; metaphors migrate across centuries; narrative structures recur; rhetorical forms are inherited; quotations survive their original contexts; idioms enter writing without announcing their genealogies. Even the most recognisable individual style is assembled within linguistic possibilities that existed before the individual and will continue afterwards. The Author therefore occupies a paradoxical position: celebrated as the origin of a work whose basic material can never originate with the author. This does not render originality meaningless. Originality can instead be understood relationally—as an unusual selection, collision, transformation or arrangement of inherited elements rather than creation from absolute nothingness. But once this concession is made, sovereign authorship begins to weaken. The writer becomes less analogous to a deity creating a world ex nihilo and more like a temporary organiser within a cultural environment already saturated with forms. The problem becomes even clearer when editors, translators, publishers, designers and readers are admitted into the picture. A manuscript may be radically shortened by an editor, transformed by translation, renamed by a publisher, visually repositioned by a cover and interpreted in ways the writer neither intended nor anticipated. Which of these interventions belongs outside authorship, and why? The apparent unity of the Author depends upon hiding much of this distributed labour behind the signature. The name on the cover compresses a network into a person. This compression has practical value, but theoretical danger begins when convenience is mistaken for ontology. The signed name encourages culture to imagine that the work possesses a single centre because it possesses a single prominently displayed origin. The Zen analogy therefore becomes more than decorative provocation. Attachment to the Author resembles attachment to the master because both substitute a reassuring centre for a more difficult encounter. Instead of confronting the instability of language, the reader can ask what the writer intended. Instead of accepting that meaning emerges through relations among text, cultural memory and reader, criticism can return to the person whose name promises an answer. The Author becomes dangerous not because authors possess too much actual power, but because belief in authorial sovereignty can terminate interpretation prematurely.

Barthes’s intervention becomes intelligible within this genealogy as an act of iconoclasm. What had to die was not the breathing person holding the pen but the theological structure that placed that person above the text. This is why the metaphor of killing the master clarifies something easily obscured by the familiarity of the phrase “death of the author”. Death here is methodological emancipation. The text must be permitted to exceed the consciousness that participated in producing it. Once released, it enters relations its writer cannot completely govern: historical circumstances change; words acquire new connotations; forgotten references become visible; readers from different cultures recognise structures unavailable to the original audience; later works retrospectively alter the significance of earlier ones. The writer cannot stand beside every future reader correcting these transformations. Nor would such correction necessarily settle the matter. An author can explain what was intended without exhausting what was produced. The distinction is crucial because it separates historical evidence from interpretative command. A letter stating what a novelist intended may be invaluable evidence, but evidence is not sovereignty. The Author dies when intention loses its status as the supreme court of meaning. Yet the Zen structure introduces a further complication that will become essential in the age of AI: killing one master can merely prepare the enthronement of another. Once the Author is removed, the Text itself can become sovereign; the Reader can become sovereign; discourse can become sovereign; later, the algorithm or Model can be imagined as the new privileged source of linguistic production. The history of authorship therefore cannot be represented as a simple liberation in which an oppressive Author disappears and decentralised meaning permanently triumphs. Cultural systems repeatedly reconstruct centres because centres make complexity manageable. They allow attribution, ownership, classification, reputation and responsibility to be organised. The critical task must consequently be repeated whenever one explanatory figure begins absorbing functions that properly belong to a network. If the writer is treated as the complete origin, kill the Author. If the reader is treated as the unlimited sovereign of meaning, interrogate the Reader. If the Model is treated as the autonomous creator merely because it generated the sentences, the same operation becomes necessary again. The road in the article’s title is therefore historical as much as metaphorical. Along it appear successive candidates for textual mastery, each promising to answer the question of where language finally begins. The significance of Barthes lies not in having performed the final execution but in revealing why the execution can never be final. The Author who is killed as metaphysical origin will shortly return in another form—as function, signature, institution and eventually computational problem. Foucault’s decisive contribution begins precisely at this point: not by resurrecting the sovereign writer, but by asking why culture continues to require something called an author after sovereignty has supposedly been destroyed.

Part II: The Corpse That Kept Writing — Barthes, Foucault and the Return of the Author

“The author is therefore the ideological figure by which one marks the manner in which we fear the proliferation of meaning.”

— Michel Foucault (1969), “What Is an Author?”

The most revealing consequence of Barthes’s execution of the Author is that the corpse continued to perform cultural work. The celebrated death did not eliminate writers, signatures, reputations or attribution; more importantly, it did not eliminate the mechanisms through which texts are gathered around names and names are transformed into principles of interpretation. Barthes’s intervention was directed against a particular hierarchy in which the writer stood chronologically and semantically before the work, as though the text were the delayed transcription of something already complete inside an individual consciousness. The Scriptor interrupts this hierarchy. Rather than preceding language as its sovereign origin, the Scriptor enters language that already exists, combining cultural codes, inherited forms, remembered phrases and circulating discourses. Writing consequently ceases to resemble the transmission of a private substance from consciousness to page. The apparent originality of the Author gives way to an activity of arrangement in which no sentence can be completely separated from the linguistic histories that make it possible. Yet the radicalism of this displacement contains its own temptation. If meaning is liberated from authorial intention, it becomes attractive to place sovereignty elsewhere, and the Reader appears ready to inherit the abandoned throne. The famous movement from the death of the Author towards the birth of the reader can therefore be misunderstood as a simple reversal of power: first the writer controlled meaning; now the reader does. Such a reversal merely changes the identity of the sovereign. If every interpretation becomes legitimate because it occurs in reading, the text risks becoming no less subordinate to the Reader than it had previously been to the Author. The death of the reader in the title of this article therefore names a second decentralisation. The reader is indispensable to the actualisation of textual possibilities but cannot plausibly function as their unlimited creator. Language constrains interpretation even while permitting plurality; genre establishes expectations; historical contexts delimit plausible references; textual structures resist certain readings more strongly than others. Neither Author nor Reader can consequently possess the work completely. The text occurs within a field of relations in which production and reception are both necessary but neither is sovereign. This double dethronement prepares the ground for a more difficult understanding of authorship. Once the writer can no longer dictate meaning and the reader cannot simply manufacture it without constraint, the question shifts from who owns the text towards which functions are being performed when culture attributes a work to someone. At precisely this point Foucault approaches the body Barthes has theoretically left behind and asks the question that prevents the funeral from becoming an ending: what remains culturally operative after the sovereign Author has supposedly disappeared?

Foucault’s author-function transforms the problem because it refuses to treat the author merely as a person standing behind discourse. An author’s name does things. It groups certain texts together while excluding others; permits relationships of consistency, development and contradiction to be constructed among works; affects the status with which discourse circulates; connects statements to regimes of ownership and responsibility; and allows culture to distinguish ordinary utterances from works regarded as possessing a particular kind of significance. “Shakespeare”, “Kafka” or “Foucault” therefore denotes an empirical individual, but the name simultaneously operates as an organising mechanism inside culture. This explains why the Author can survive his own theoretical death. Sovereign intention may be rejected while the classificatory machinery associated with authorship remains indispensable. A library still requires names; publishers require contracts; universities require attribution; copyright requires identifiable relations between works and rights-holders; journalism requires responsibility; literary markets require reputations. Genette’s account of paratext further demonstrates that even the apparently simple encounter between Author and Reader is surrounded by mediating structures: titles, subtitles, prefaces, covers, illustrations, blurbs, typographical decisions and other thresholds frame the work before interpretation begins. Bourdieu widens the field again. Cultural value emerges within relations among writers, publishers, critics, institutions, prizes, educational systems and competing forms of symbolic capital. Sapiro similarly situates authorship within social structures that determine not only how writers are recognised but how literary production becomes a profession, an identity and a position within institutional life. The Author consequently returns not as the metaphysical genius Barthes attempted to destroy but as a distributed cultural necessity. This distinction is decisive. One can reject the proposition that the author possesses the final meaning of a work while continuing to require authorship for attribution and responsibility. One can acknowledge collaborative production while preserving the usefulness of a name on the cover. One can recognise editorial intervention without pretending that editor and writer occupy identical positions. The problem begins when these different functions are compressed once again into a single sovereign figure. The name then appears to explain what it merely organises. The Author becomes a convenient interface through which a complicated cultural network is represented as an individual. In this sense, Foucault does not simply resurrect what Barthes killed. He reveals why the corpse cannot be discarded. The author-function is a technology for managing discourse. It restricts the potentially unlimited circulation of meaning by establishing boundaries around textual bodies and connecting those bodies to identities. This is why the Author continually returns: not because Barthes failed to kill him theoretically, but because institutions repeatedly rebuild him operationally.

The distinction between Author and author-function becomes extraordinarily important once generative AI enters textual production. The Model can perform activities previously treated as signs of authorship without automatically acquiring the social functions historically attached to an author’s name. It can generate a title, compose paragraphs, reorganise an argument, simulate a genre, propose imagery, summarise research and rewrite a text in another register. If authorship were defined exclusively by the production of linguistic sequences, such capacities would make the conclusion almost unavoidable: the machine writes, therefore the machine is an author. Foucault’s framework demonstrates why the inference is premature. Production is only one component of the cultural structure condensed into authorship. The Model does not automatically decide that a work should be published, establish the institutional conditions under which it circulates, acquire a professional reputation through the consequences of its statements, or bear responsibility in the same manner as the person or organisation releasing those statements. Yet the opposite conclusion—that AI is merely a neutral instrument—is equally inadequate. A typewriter does not propose a different argument when the writer presses a key. A word processor does not spontaneously generate three alternative conclusions, imitate an established voice or transform a rough instruction into a coherent essay. Generative AI occupies an unfamiliar intermediate position because it can exercise substantial linguistic agency without automatically possessing authorial sovereignty. This fractures the author-function internally. The entity producing sentences, the person establishing intention, the agent selecting among alternatives, the institution publishing the work and the subject accepting responsibility may now be different participants. The Prompter emerges from this separation, but even the Prompter cannot simply be crowned the successor to the Author. To do so would merely relocate sovereignty from the hand that writes to the hand that instructs. The same applies to the Reader and the Model. Each can become another master encountered on the road. The history running from Barthes through Foucault therefore reveals something more consequential than the death and return of a literary category. It reveals authorship as a recurrent act of compression: culture gathers heterogeneous operations and places them beneath a name because a name makes complexity governable. Generative AI now forces those operations apart with unprecedented visibility. The corpse that kept writing has begun to fragment. Intention can occupy one location, formulation another, editing a third, publication a fourth and interpretation innumerable others. The next theoretical task is therefore not to determine which participant deserves possession of the old title. It is to ask whether authorship can continue after the disappearance of the centre that once made the title appear singular.

Part III: Kill Him Again — AI, the Prompter and Authorship Without a Centre

“Language is not an instrument that man possesses, but an event that possesses man.”

— Maurice Blanchot (1969), The Infinite Conversation

Generative artificial intelligence does not resolve the problem left by Barthes and Foucault; it makes that problem materially visible by separating operations that print culture could conveniently gather beneath the name of a single writer. When a person begins with an idea, formulates a prompt, receives several machine-generated versions, rejects most of them, rewrites another, requests a different structure, restores a deleted irregularity, selects an image and finally publishes the result, the apparently elementary question “Who wrote this?” begins to malfunction. The question assumes that writing is one activity and therefore expects one agent. In practice, the process has already fragmented into intention, instruction, linguistic generation, comparison, revision, selection, authorisation and publication. A Model may generate most of the visible sentences while contributing nothing to the original decision that those sentences should exist. Conversely, a human may type relatively few words while determining the subject, intellectual direction, constraints, acceptable evidence and final form. Neither word count nor keystroke ownership can adequately describe this relationship. The Prompter initially appears to offer a solution because the prompt restores human intentionality to the centre: the machine answers because somebody asks. Yet this solution becomes another version of the old Author as soon as prompting is treated as sovereign creation. A prompt does not contain its output in the manner in which a blueprint contains specified dimensions. It opens a probabilistic field in which the Model produces possibilities that may exceed, distort, simplify or unexpectedly improve the initiating instruction. The Prompter consequently governs neither every linguistic decision nor every semantic consequence. Nor does the Model. It generates language from learned statistical relationships without thereby becoming the sole source of the cultural materials, communicative purpose or institutional conditions through which that language acquires significance. The temptation to declare either participant the “real author” therefore repeats precisely the intellectual gesture this article has been questioning: confronted by a distributed process, culture searches for a master. First the writer possessed the text; then the Scriptor seemed to possess recombination; the Reader appeared to inherit meaning; the Prompter seemed to possess intention; now the Model appears to possess production. Each claim isolates one genuine function and mistakes it for sovereignty over the whole. AI makes this mistake increasingly difficult to sustain because its practical operation exposes the distance between producing language and authoring the conditions under which language becomes a work. If the Prompter is enthroned because intention matters, the old Author returns through instruction. If the Model is enthroned because linguistic execution matters, the old Author returns through computation. Both must therefore be “killed” again—not eliminated from the process, but deprived of the fantasy that one position contains the entire causal and semantic origin of the text.

What emerges after this second execution can be described as authorship without a centre. The formulation does not imply that texts arise mysteriously without causes or that all participants become interchangeable. It means that the functions historically compressed into authorship can occupy different locations without any one of those locations becoming sufficient to explain the finished work. A contemporary human–AI text may begin in a human dissatisfaction, memory, question or conceptual intuition. Its factual material may originate in interviews, archives, books or databases produced by innumerable other people. Its first linguistic architecture may be generated computationally. Its strongest sentence may appear unexpectedly during machine generation; its weakest argument may survive because a human fails to detect it. Subsequent prompts may restructure the text; human revision may restore an idiosyncrasy erased by optimisation; an editor may remove passages accepted by both Prompter and Model; a publisher may impose a title; a platform may determine visibility; readers may establish associations none of the preceding agents anticipated. Authorship under these conditions resembles a network more closely than a pyramid. Yet the network metaphor must itself be used carefully because distribution does not mean equality. Different participants possess different kinds of agency. The human can have biographical stakes in the argument. Sources provide information without necessarily controlling its eventual use. The Model can exercise enormous linguistic agency while lacking an independently established reason for producing this particular work. Editors can alter form while remaining institutionally invisible. Readers can transform meaning without retroactively producing the sentences they interpret. The conceptual advantage of authorship without a centre is precisely that it allows these asymmetries to remain visible. It does not solve complexity by declaring everyone a co-author. Such inflation would make authorship meaningless. Instead, it treats authorship as a configuration of relations whose components must be identified separately. This also reveals why the question of originality changes in the generative environment. If originality is defined as language untouched by preceding language, neither human nor machine can satisfy the criterion. If it is defined instead as the emergence of a significant configuration from inherited materials, originality can occur at several levels: in the initial problem, in the relation established between sources, in an unexpected generated formulation, in selection among alternatives, in deliberate rejection of optimisation or in the architecture of the final assemblage. The location of originality can therefore migrate within the process. AI does not abolish originality; it makes increasingly implausible the assumption that originality must correspond to one identifiable moment in which a sovereign mind creates linguistic form from nothing.

The dissolution of the centre nevertheless encounters a limit that prevents decentralisation from becoming an alibi: responsibility cannot be allowed to disappear merely because production has become distributed. Indeed, generative AI makes the separation between authorship and accountability more urgent. A newspaper cannot reasonably answer a fabricated quotation by claiming that no single agent authored the final sentence. A scholar cannot escape responsibility for invented references because a Model supplied them. A publisher cannot dissolve accountability into an algorithmic network after choosing to circulate a generated text. The metaphysical Author may be unnecessary, but responsible human and institutional actors remain indispensable. This distinction provides the point at which the article’s apparently destructive argument becomes constructive. Killing the sovereign Author does not require killing attribution, intellectual honesty, ownership or responsibility. It requires preventing these practical necessities from being converted into a theory of absolute textual origin. The person whose name appears on a work may increasingly function not as the individual who physically generated every word but as the person who authorises the configuration and accepts responsibility for its publication. This transforms the signature. Instead of meaning “every linguistic element originated inside me”, it may increasingly mean “I accept this arrangement of human and machine operations as the work for which I answer”. Such a definition is particularly compatible with the Delegated Author, but authorship without a centre pushes beyond even that figure. Delegation still implies a relatively clear direction from principal to agent; future creative systems may involve iterative relationships in which ideas arise through interaction and cannot easily be assigned to one side of the exchange. The Prompter may begin with one intention and discover another through machine responses. The Model may produce an accidental formulation that redirects the project. The human may then recognise significance that was neither explicitly requested nor computationally intended. Creative causality becomes recursive. Under such conditions, searching for the “real author” can become less illuminating than reconstructing the sequence of transformations through which the work emerged. This is the final significance of the command to kill the Author on the road. The road does not end with Barthes, and the execution cannot happen only once. Whenever a new technology, theory or cultural practice identifies another figure as the supreme origin of meaning, the idol has returned. The Scriptor cannot occupy the throne. The Reader cannot occupy it. The Prompter cannot occupy it. The Model cannot occupy it. What remains after these repeated dethronements is not a world without human creativity but a more demanding account of it: language produced through relations, originality emerging through transformation, meaning without a single proprietor, and responsibility preserved without the mythology of sovereignty. The most radical consequence of generative AI may therefore be neither the death of the writer nor the birth of the machine author. It may be the moment when authorship finally ceases to require a king.

Conclusion: Nobody Wrote This Alone

“Where there is a work, there is no author.”

— Maurice Blanchot (1955), The Space of Literature

The history traced through this article suggests that the death of the Author was never an event that could happen once. It is a recurring intellectual operation required whenever one participant in textual production is elevated into the sovereign explanation of the whole. The empirical writer survives this operation perfectly well. Human beings continue to imagine, research, remember, formulate intentions, arrange materials, compose sentences, revise drafts and attach their names to works. What disappears is the stronger proposition that these activities establish a single centre from which the text receives its complete origin and final meaning. Barthes attacked that proposition by transforming the Author into the Scriptor and locating writing within an already existing universe of language. Foucault demonstrated why the resulting disappearance could never be complete: societies continue to require mechanisms of attribution, classification, ownership and responsibility, and the author-function supplies them. The Reader subsequently appeared as another possible centre, but interpretation cannot be made sovereign without reproducing the structure it supposedly escaped; texts permit plurality while continuing to resist infinitely arbitrary meanings. Generative AI now performs the most dramatic material fragmentation of this genealogy. Intention can remain human while linguistic execution becomes computational; machine production can be followed by human selection; editorial transformation can be distributed across both; publication can belong to an institution; interpretation proliferates among readers who may never know which sentences originated from which operation. The traditional question “Who wrote this?” survives, but it increasingly asks several questions while pretending to ask one. Who conceived it? Who supplied its knowledge? Who formulated its sentences? Who selected among alternatives? Who revised them? Who decided that the work was finished? Who placed a name upon it? Who permitted its publication? Who answers when it is false, harmful or fraudulent? These questions need not produce the same name. That is the significance of authorship without a centre. It does not describe the disappearance of agency but its differentiation. Nor does it declare every contributor an author, which would merely dissolve the term into uselessness. It asks instead that the heterogeneous operations hidden inside authorship be recognised before culture once again compresses them into another reassuring sovereign. The writer was never alone; the Reader was never omnipotent; the Prompter does not contain the Model’s output; the Model does not contain the cultural archive from which its capacities emerge. Nobody wrote this alone—not because nobody wrote it, but because writing has never been as solitary as the mythology of the Author required.

The command in the title can therefore be understood finally as a rule against intellectual idolatry. If the Author is encountered as the absolute origin of the work, kill him. If the Reader appears as the unrestricted creator of meaning, dethrone the Reader. If the Prompter is celebrated as a new genius because a sufficiently ingenious instruction generated an impressive result, refuse the coronation. If the Model is proclaimed the new Author because it physically produced the sentences, repeat the operation again. None of these acts requires denying what each participant actually contributes. On the contrary, removing sovereignty allows those contributions to be described more accurately. The human can supply intention without possessing every formulation; the Model can generate language without possessing the complete purpose of the work; the Reader can create interpretations without producing the object interpreted; institutions can confer legitimacy without becoming the origin of meaning. What cannot be permitted to dissolve with the sovereign Author, however, is responsibility. The disappearance of a single centre of textual causation does not justify the disappearance of accountable actors. When a human chooses to publish machine-generated language, when an editor accepts it, when a newspaper circulates it or when a scholar places a name beneath it, practical responsibility remains even if metaphysical authorship has become distributed. The signature may therefore acquire a new significance in the generative age. Instead of promising that every word originated inside the consciousness whose name appears beneath it, the signature can increasingly function as an act of authorisation: this is the configuration of sources, inherited language, human decisions and computational operations that I have chosen to release, and for whose release I accept responsibility. Such a transformation would preserve what remains socially necessary about authorship while abandoning what has become theoretically untenable. The road therefore continues beyond the death of the Author, beyond the death of the sovereign Reader and beyond the appearance of the Prompter and the Model. There will almost certainly be another master waiting further along it, another technology or theory promising to reveal where language truly originates. The lesson is not that creators must disappear but that no creator should become the final explanation of creation. Authorship survives most convincingly when sovereignty does not. The writer remains, the Model remains, the Reader remains, and responsibility remains; only the throne must remain empty.

The Author disappears not when machines learn to write, but when writing can no longer be credibly returned to a single sovereign origin. What remains is a shifting constellation of intention, language, memory, machine, institution and reader, none of which can claim the text entirely as its own. The task of the future is therefore not to discover who should inherit the Author’s throne, but to preserve responsibility while refusing to place anyone upon it. If you meet an Author on the road, kill him—because the work can begin to belong to language only when its supposed master finally steps aside.

“The work requires of the writer that he lose everything he might construe as his own ‘nature’.”

— Maurice Blanchot (1955), The Space of Literature

References

Ayolov, P. (2026) The Conspiracy of Speech. Vol. I of The Planned Obsolescence of Language: Miscommunication Trilogy.

Ayolov, P. (2026) The Entropy of Communication. Vol. II of The Planned Obsolescence of Language: Miscommunication Trilogy.

Ayolov, P. (2026) Negentropic Culture: Against the Entropy of Words. Vol. II, Part 2 of The Planned Obsolescence of Language: Miscommunication Trilogy.

Ayolov, P. (2026) The Tower of Babble. Vol. III of The Planned Obsolescence of Language: Miscommunication Trilogy.

Ayolov, P. (2026) ‘The Death of the Reader 2.0: From Roland Barthes’ Scriptor to the AI Prompter’.

Ayolov, P. (2026) ‘The Delegated Author: Generative AI and the Fragmentation of Authorship’.

Ayolov, P. (2026) ‘Language as a Tissue of Citations: How a Self-Quoting Mechanism Conquered the Human Mind’.

Ayolov, P. (2026) ‘Automating the Wilde–Bayard Paradox: LLMs, Predictable Equilibrium and the Death of Deep Listening’.

Ayolov, P. (2026) ‘The Neo-Awakening of Information: The Illusion of Self, LLMs, and the Preservation of the Immortal Pattern’.

Ayolov, P. (2026) ‘From Snubbing to Znubbing: The Amputation of Attention’.

Barthes, R. (1977 [1967]) ‘The Death of the Author’, in Heath, S. (trans.) Image–Music–Text. New York: Hill and Wang, pp. 142–148.

Barthes, R. (1974) S/Z. Translated by R. Miller. New York: Hill and Wang.

Benjamin, W. (2008 [1936]) The Work of Art in the Age of Its Technological Reproducibility and Other Writings on Media. Cambridge, MA: Harvard University Press.

Blanchot, M. (1982 [1955]) The Space of Literature. Translated by A. Smock. Lincoln, NE: University of Nebraska Press.

Blanchot, M. (1993 [1969]) The Infinite Conversation. Translated by S. Hanson. Minneapolis: University of Minnesota Press.

Boden, M. A. (2004) The Creative Mind: Myths and Mechanisms. 2nd edn. London: Routledge.

Bourdieu, P. (1993) The Field of Cultural Production: Essays on Art and Literature. Edited and introduced by R. Johnson. Cambridge: Polity Press.

Bourdieu, P. (1996) The Rules of Art: Genesis and Structure of the Literary Field. Translated by S. Emanuel. Stanford, CA: Stanford University Press.

Burke, S. (1992) The Death and Return of the Author: Criticism and Subjectivity in Barthes, Foucault and Derrida. Edinburgh: Edinburgh University Press.

Chartier, R. (1994) The Order of Books: Readers, Authors, and Libraries in Europe between the Fourteenth and Eighteenth Centuries. Translated by L. G. Cochrane. Stanford, CA: Stanford University Press.

Derrida, J. (1976 [1967]) Of Grammatology. Translated by G. C. Spivak. Baltimore, MD: Johns Hopkins University Press.

Eco, U. (1979) The Role of the Reader: Explorations in the Semiotics of Texts. Bloomington: Indiana University Press.

Eco, U. (1990) The Limits of Interpretation. Bloomington: Indiana University Press.

Eco, U. (1992) Interpretation and Overinterpretation. Cambridge: Cambridge University Press.

Fish, S. (1980) Is There a Text in This Class? The Authority of Interpretive Communities. Cambridge, MA: Harvard University Press.

Floridi, L. and Chiriatti, M. (2020) ‘GPT-3: Its nature, scope, limits, and consequences’, Minds and Machines, 30, pp. 681–694.

Foucault, M. (1977 [1969]) ‘What Is an Author?’, in Bouchard, D. F. (ed.) Language, Counter-Memory, Practice: Selected Essays and Interviews. Translated by D. F. Bouchard and S. Simon. Ithaca, NY: Cornell University Press, pp. 113–138.

Genette, G. (1997) Paratexts: Thresholds of Interpretation. Translated by J. E. Lewin. Cambridge: Cambridge University Press.

Hayles, N. K. (1999) How We Became Posthuman: Virtual Bodies in Cybernetics, Literature, and Informatics. Chicago: University of Chicago Press.

Hayles, N. K. (2005) My Mother Was a Computer: Digital Subjects and Literary Texts. Chicago: University of Chicago Press.

Iser, W. (1978) The Act of Reading: A Theory of Aesthetic Response. Baltimore, MD: Johns Hopkins University Press.

Jenkins, H. (2006) Convergence Culture: Where Old and New Media Collide. New York: New York University Press.

Linji Yixuan (2009) The Record of Linji. Translated by R. F. Sasaki and edited by T. Kirchner. Honolulu: University of Hawai‘i Press.

Manovich, L. (2001) The Language of New Media. Cambridge, MA: MIT Press.

Manovich, L. (2013) Software Takes Command. New York: Bloomsbury Academic.

McLuhan, M. (1962) The Gutenberg Galaxy: The Making of Typographic Man. Toronto: University of Toronto Press.

McLuhan, M. (1964) Understanding Media: The Extensions of Man. New York: McGraw-Hill.

Sapiro, G. (2016) The Sociology of Literature. Translated by M. Ryan. Stanford, CA: Stanford University Press.

Shannon, C. E. (1948) ‘A mathematical theory of communication’, Bell System Technical Journal, 27, pp. 379–423 and 623–656.

Turing, A. M. (1950) ‘Computing machinery and intelligence’, Mind, 59(236), pp. 433–460.

Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, Ł. and Polosukhin, I. (2017) ‘Attention is all you need’, Advances in Neural Information Processing Systems, 30, pp. 5998–6008.

Wiener, N. (1948) Cybernetics: Or Control and Communication in the Animal and the Machine. New York: John Wiley & Sons.