(Adapted for this entry from the chapter of the same name in Augere: The Death of the Author)

By Lázara Ávila Fernández

This text forms part of the broader philosophical framework developed in Augere: The Death of the Author. It argues that the decisive question raised by language models is not authorship itself, but the expropriation of the human act of signification.

At this unique moment in history, many of us are asking ourselves this question. The advent of language models—or large language models (LLMs)—has transformed the landscape of creative writing to such an extent that it is difficult to determine with certainty whether a given text was written exclusively by a human being or not.

In my personal view, I believe in writing as the result of the conscious being that is the individual—the conscious agent who observes a phenomenon and is capable of infusing art into words and transporting others to unimaginable places, societies, and worlds. An act that takes shape through the materiality of the sign, but which, without the productive and creative consciousness of the human, ceases to make sense. And I do not believe we have created the figure of the writer to understand writing, because both the writer and the act of writing are inextricably linked, and one cannot exist without the other. And I am referring to the craft, to the act of thought—not to a mere copy, and certainly not to algorithmic language, which may fulfill its operational function but never an ethical or social one. I am referring to the two parallel strands that make up that act: the first, the material or subjective imprint; the second, the conscious intention preceded by the interpretant. Authorship, the act of authorship—not the modern institutionalization of that act—is geographically universal and genuine. There was a creative act in Homer, in the Persian rhapsodes, in the oral traditions not registered with any copyright office, in the caveman who left his mark on the rock, and in the civilizations of the Aztecs and the Maya. The institution of the author is European and modern; the act belongs to the human species.

Today, when some scholars argue that the figure of the author or writing itself as an act has disappeared, become extinct, or is in the process of being lost, it is worth asking whether they are viewing the current phenomenon as a whole or only in part. Roland Barthes, in his famous essay The Death of the Author, argued that a text does not have a single meaning dictated by the intention of its creator, but rather that true meaning comes to life and becomes multifaceted in the mind of the reader. You may or may not agree with this approach. However, it is worth reading between the lines: meaning that comes to life and becomes multifaceted in the mind of the reader. On the one hand, it speaks of the act of signification, and on the other, of intention—and none of these concepts is the same. Without someone who has given meaning—who has previously carried out the act of signification and transduced it into the written word or any other form of expression—there can likely be no reading.

More recently, David Gunkel has argued that, although Barthes never encountered a language model, his essay accurately anticipated our current situation: language models produce written content without a living voice to animate and authorize their words, so that the text remains, literally, unauthorized.

This means, as I said before, that the debate is not only taking place in the public sphere but has also reached academia. Thus, for example, we find the academic article “Language Models’ Hall of Mirrors Problem: Why AI Alignment Requires Peircean Semiosis” by David Manheim, who, in a separate paragraph at the end, highlights which parts of his own article were generated by this or that language model. Similarly, in 2025, a U.S. appeals court denied authorship to a work created autonomously by AI, a decision that the Supreme Court upheld in 2026.

The picture could thus be interpreted as apocalyptic: the figure of the author—and therefore writing itself—in irreversible decline.

But is this a correct interpretation of our reality and of what the act of writing and signifying truly represents? First of all, let’s make a key distinction: I am not referring to “copying and pasting” what the algorithm generates. That makes no sense to me—not even in a Barthesian sense. Barthes spoke of intention, and the algorithm lacks it; so does the person who copies and pastes. But my argument is not about intention alone—it is about the entire act of signification, which the model simply does not perform.

The question is poorly phrased. Should the output be accepted as a complete sign? Has the figure of the author truly died? Has the criterion—writing—truly died? Are we going to be content from now on with only the figure of the reader and take it for granted that the language model is the one who signs what we read?

To do so would be to lose sight of, among other factors, the context window—where interaction with the language model occurs—and thus the constraint that the human—the sad and vaunted logos—exerts on that window and, consequently, on the algorithmic vector. It would also mean ignoring the expropriation of the act of signification that occurs from the very moment human meaning is treated as a resource and used to train language models. This expropriation is not limited to that single moment; it manifests itself even when we copy and paste the output generated by the model without realizing that we are “validating” a quasi-sign—one to which we have not given our world. The explanation is much more extensive; for the sake of brevity, I’ll stop here. I’ll leave you with just one thought: no, the figure of the author is not dead; if it were to die, we would be ceding invaluable ground to the machine. Language models are not human, and they do not mean anything—they merely calculate. Inside the black box, only correlations and matrix operations take place. Inside our brains, interpretations of the world around us occur, from which we interpret, think about the world, and create.

Today, after reading this, when you interact with the language model again—whether you’re a business owner devising a marketing strategy, a homemaker looking up a recipe, or a writer—pause for a second and keep in mind that the language model returns quasi-signs that are nothing without your brain. They are nothing without your world.

Note: This English version is a translation of the original Spanish text, which appears in Mujer Bienestar, Volumen III (a literary project by Hispanic women on diverse topics) and was adapted from the chapter of the same name in the author's book Augere: The Death of the Author. The translation was produced using DeepL, triangulated with Claude and DeepSeek, and the final version was reviewed by the author.