A recent controversy surrounding an award-winning short story, suspected by some readers and commentators of being generated or heavily assisted by artificial intelligence, has been treated as a warning to writers. If literary judges and publishers can no longer reliably distinguish human from machine-produced prose, then perhaps authorship itself is in danger. Perhaps the author is about to become one more occupation displaced by automation.
That conclusion moves too quickly. These developments do not immediately threaten authorship as such. They challenge a view of authorship formed under particular economic and market conditions. For a long time, those conditions allowed us to treat the visible signs of authorship as though they were authorship itself. A polished manuscript, a professional cover, an established imprint, a prize, a favourable review, or a place on a bookstore shelf appeared to confirm the presence of an author. In many cases, they did. Yet they did so indirectly. They were signs produced by a system in which competent writing, editing, printing, distribution, and public recognition were comparatively scarce and expensive.
The market needed authorship to be legible at a glance. Readers could not investigate the full history of every book they encountered. Publishers could not explain every editorial decision. Booksellers could not place the writer’s notebooks, discarded drafts, conversations, research, and years of experience beside each finished volume. Instead, the system developed proxies. The quality of the prose, the reputation of the publisher, the endorsements on the cover, the price, the sales record, and the institutional recognition surrounding a book all helped readers decide whether it deserved attention.
These signs were never meaningless. Under earlier conditions of production, they often pointed toward genuine labour, acquired skill, sustained thought, and editorial care. The difficulty is that we gradually allowed the signs to carry more weight than the realities they represented. A book that looked and sounded professionally produced could be received as the work of an author before the reader knew anything about the person who stood behind it. Economic exchange favoured this simplification. The book had to circulate as a product, and the author often appeared primarily as the producer or brand attached to that product.
Artificial intelligence disrupts the connection between the sign and the reality. It can produce fluent sentences, coherent structure, an attractive title, persuasive promotional language, and a plausible imitation of literary style with remarkable speed. Human users can then revise the output, remove its awkwardness, add personal details, and shape it toward an identifiable voice. The resulting work may display many of the signs that once helped readers infer sustained human authorship. Yet those signs can now be reproduced without the history of attention, struggle, memory, judgment, and revision that readers assumed they represented.
Walter Benjamin’s account of the artwork’s “aura” helps explain why this disruption feels larger than the arrival of another writing tool. Benjamin associated aura with the unique existence of a work within a particular history, tradition, time, and place. Mechanical reproduction did not simply make additional copies available. It altered the relationship between the cultural object and the singular history from which its authority had arisen.
Generative artificial intelligence intensifies this disturbance. It does not merely reproduce a finished work. It can reproduce the signs through which readers once inferred that a singular history of human attention stood behind the work. Polish, stylistic coherence, professional presentation, and even the appearance of a recognizable voice can now be generated without the corresponding history of labour and experience. Artificial intelligence therefore does not immediately eliminate authorship. It weakens the aura through which modern markets and institutions taught us to recognize it.
This is why the failure of detection feels so unsettling. The problem is larger than the inability of a judge or software program to identify machine-generated prose. Detection belongs to the same system of signs. It seeks another visible feature that will allow an institution to sort works into clean categories: human or machine, permitted or prohibited, authentic or fraudulent. That sorting may remain necessary for contracts, copyright, competitions, and disclosure policies. It cannot provide a complete account of authorship.
When polish becomes abundant, polish ceases to be reliable evidence of human presence. This does not make craftsmanship irrelevant. Readers still deserve sentences that have been considered, structures that clarify rather than confuse, and language appropriate to the work. A badly written book remains difficult to read. Yet technical competence can no longer bear the philosophical and moral weight we often placed upon it. It may tell us that a text is effective. It cannot, by itself, tell us who is speaking, what relationship that person has to the words, or whether anyone is prepared to answer for them.
A renewed account of authorship must therefore distinguish style from voice. Style consists partly of recognizable verbal patterns: vocabulary, rhythm, imagery, sentence length, tone, and preferred forms of argument. These patterns can be studied and imitated. A machine can produce a convincing approximation of them, just as a human writer can imitate the style of another writer.
Voice is more difficult to define because it belongs to a life rather than only to a text. A voice carries memory, education, loyalties, injuries, relationships, convictions, failures, and changes of mind. It emerges from encounters with other people and from the particular world in which a person has learned to speak. The writer’s voice may be disciplined and carefully composed, but it remains connected to someone who can be questioned, contradicted, praised, blamed, corrected, or forgiven.
A machine can imitate the verbal surface of a voice. It cannot occupy the life from which that voice arose. It cannot remember the conversation that altered an argument, feel embarrassment over a careless claim, reconcile with someone harmed by its words, or return years later to revise a conviction. It cannot receive the consequences of publication as events within a life. These absences do not mean that every human sentence is profound or morally serious. They mean that human authorship includes possibilities of responsibility that textual style alone cannot display.
Hannah Arendt offers language for understanding why a voice cannot finally be separated from a life. In The Human Condition, speech and action are the means through which a person appears among others. Our deeds reveal what has happened, but speech helps disclose who has acted, what the action meant, and how the actor understands what has been begun. Neither speech nor action is entirely controlled by the person from whom it proceeds. Their meaning unfolds within a world shared with others.
This disclosure of “who” someone is cannot be reduced to a list of qualities or stylistic traits. It emerges across time through the relationship between words and deeds. A person may imitate a style or adopt a public vocabulary, but a recognizable identity becomes visible through promises kept or broken, judgments defended or revised, responsibilities accepted, and actions undertaken before others. Authorship, understood in this Arendtian sense, belongs to the appearance of a person within a shared world.
This is where authorship becomes a form of answerability. To author a work is more than to cause a sequence of words to appear. It is to establish and maintain a relationship with those words. The author can explain why a passage was written, which experiences informed it, what sources changed the argument, where uncertainty remains, and why one formulation was chosen over another. The author can hear criticism, acknowledge error, defend a judgment, revise the work, or admit that the finished text says more than was intended.
Answerability does not require omniscience. Authors regularly misunderstand their own motives, overlook implications, and discover meanings in their work through the responses of readers. Responsibility is possible precisely because the work enters a shared world where the author does not control its reception. To publish is to release words among other people while remaining connected to what has been released.
This account also helps us avoid treating technological purity as the sole measure of authorship. A writer has always depended upon materials, conventions, teachers, editors, sources, conversations, and inherited forms of language. The use of a tool does not automatically sever the relationship between a writer and a work. At the same time, describing artificial intelligence as merely another tool can conceal important differences in the scale and kind of contribution it makes. A spell-checker corrects a surface error. A generative system can propose arguments, images, structures, transitions, and entire passages. The more judgment a writer delegates, the more seriously the writer must ask what remains genuinely theirs to answer for.
The central question is therefore neither whether a perfectly pure human process can be proven nor whether any use of artificial intelligence invalidates a work. The stronger question concerns the continuity of human responsibility. Did the writer remain intellectually present? Were the claims understood and judged rather than merely accepted? Can the writer truthfully describe how the work was made? Is the writer prepared to defend, revise, and bear the consequences of what appears under their name?
This renewed account also clarifies why reader trust may become increasingly important. In a marketplace crowded with competent and professionally packaged books, readers may seek more than the finished object. They may want a history of thought. They may follow essays, interviews, lectures, conversations, newsletters, and earlier books because these reveal how a writer’s ideas have developed. The relationship does not prove that every sentence was produced without technological assistance. It gives readers grounds for believing that a recognizable person has remained present across time.
Such trust should not be reduced to a marketing strategy. An audience can be manipulated, and public visibility can become another manufactured sign. Trust grows through continuity between words and conduct, through honest responses to criticism, through appropriate disclosure, and through the willingness to admit uncertainty or change. The author becomes credible through a sustained public relationship rather than through the performance of flawless certainty.
The practical implications are relatively simple. Writers should decide what kinds of assistance they are willing to use before convenience makes the decision for them. They should preserve notes, drafts, sources, revision histories, and records of development. These materials may become useful in contractual or legal disputes, but their deeper value is that they preserve the history of the work. Writers should disclose uses of artificial intelligence when those uses materially affect a reader’s understanding of what has been created. Most importantly, they should remain able to explain the judgments embodied in the finished text.
The growth of artificial intelligence may therefore expose a weakness in the market-shaped image of the author. We learned to identify authorship through scarcity, polish, packaging, institutional approval, and commercial circulation. Those signs once had a dependable relationship to human labour, though never a perfect one. That relationship is weakening. The signs can now be manufactured more quickly than the realities they were meant to represent.
Yet authorship survives this disruption because it was never reducible to those signs. A fuller account begins with a person situated within a life, formed through relationships, receiving language from others, and making judgments that enter a common world. The author stands behind the work without claiming total control over it. The author acknowledges the sources from which it came, receives its unforeseen consequences, and remains available to answer.
The arrival of machines that can generate persuasive language may eventually do more than unsettle writers. It may help us recover what the market allowed us to forget. An author is not simply the name attached to a polished product or the originating cause of a string of sentences. An author is someone whose life remains meaningfully connected to the words, someone who can say how they came to be written, why they matter, and what responsibilities follow from placing them before others.
In the age of artificial intelligence, the most human mark of authorship may be neither stylistic originality nor technological purity. It may be the willingness to remain present.
Raymond Klassen - Ideals and Identities 