The Author Was Never Alive: Symbols, Machines, and the Collective Nature of Writing

 

Introduction

Roland Barthes famously declared in 1967 that the author is dead. More than half a century later, this pronouncement has gained eerie relevance with the rise of artificial intelligence in language production. Large Language Models (LLMs), capable of composing poems, essays, and even philosophical musings, appear to fulfill Barthes’ vision of writing unbound from the authority of personal intention. But what does it mean that machines can generate coherent text without consciousness, context, or experience? Rather than mourning the human in this equation, we might ask instead: was language ever truly authored in the first place?

This essay explores the idea that both human and machine writing exist within a vast, symbolically mediated network where meaning is never original, only iterative. Derrida’s aphorism, “Il n’y a pas de hors-texte” (“There is nothing outside the text”), resonates here not only as a philosophical provocation but as a description of how language operates. Quoting Peirce, Derrida reminds us: omne symbolum de symbolo — symbols grow from symbols. In this light, intelligent software trained on global textual data does not simulate writing; it reveals its collective, distributed nature.

Barthes, Derrida, and the Fiction of the Author

In "The Death of the Author," Barthes calls into question the romantic idea of a writer as the origin of meaning. He argues, “A text is not a line of words releasing a single 'theological' meaning... but a multi-dimensional space in which a variety of writings, none of them original, blend and clash.” The figure of the Author, then, is a myth constructed to guarantee fixed interpretation.

This insight resonates with Jacques Derrida’s deconstruction of the metaphysics of presence. In Of Grammatology, he critiques the traditional Western privileging of speech over writing—a hierarchy grounded in the illusion that spoken language is closer to thought, and therefore more authentic. Drawing on Saussure’s assertion that “in language there are only differences, and no positive terms,” Derrida argues that language has no essence and no origin. It is a system of differences in which signs signify only by virtue of the traces of what they are not—through the play of absence rather than any intrinsic presence.

Peirce's semiotic theory adds another layer: symbols do not emerge from experience alone but from a chain of prior signs. Thus, the author is not a source but a point of relay within an ongoing semiotic process. Whether in human hands or algorithmic processes, texts are produced not by sovereign creators but through inherited structures.

From the Scriptor to the System: Writing in the Age of AI

Barthes proposes that the modern "scriptor" replaces the traditional author. This figure “no longer bears within him passions, humours, feelings, impressions,” but simply arranges existing language. In the case of AI, this proposition becomes literal. Contemporary text generators possess no inner life, no history, and no intentionality. Yet, their output often rivals or surpasses human prose in fluency.

This eerie fidelity to linguistic form is not accidental. Machine learning systems are trained on terabytes of written material produced across languages, cultures, and historical periods. What they offer is not invention but synthesis. As the modern essay The Author Is Dead: Barthes' Uncanny Prophecy and the Age of AI observes, AI fulfills Barthes' prediction: writing becomes “an anonymous, neutral composite” generated from a multitude of voices.

Importantly, these programs do not simulate originality; they instantiate what writing already was—a recombination of socially conditioned signs. They render explicit the logic implicit in all composition: the weaving together of traces, echoes, citations.

Rethinking the Social in Language

A common critique of machine-generated writing is that it lacks the sociality of human discourse. However, one might argue the opposite. Because these systems are trained on global, temporally expansive datasets, they arguably express the collective character of language more faithfully than any single speaker ever could.

Human perception of the world is not direct; it is symbolically mediated. Our senses do not provide pure data but are already embedded in networks of signification. Whether through gesture, image, speech, or script, all communication is filtered through semiotic systems. Software such as GPT engages with one modality (written language), but that mode is sufficient to reproduce the conditions under which meaning arises.

As Derrida notes, writing is not secondary to speech but the very condition for the deferral of meaning—the spacing and temporalizing that make thought communicable. In this sense, the difference between a human interpreter and an intelligent machine is not a matter of kind but of interface. Both participate in a symbolic continuum where meaning is always deferred, never delivered.

Writing as Collective Process, Not Genius

To reframe authorship is not to denigrate human creativity but to place it within the ecological context of symbolic systems. Genius, as traditionally conceived, rests on the illusion of isolation. But language is a system where everything holds together. No word, no idea, no expression exists in a vacuum.

Contemporary AI systems do not produce meaning in spite of this absence of originality; they do so because originality has always been a retrospective attribution. What we call creation is often reconfiguration. In this, intelligent machines mirror the human condition more than we might like to admit. They do not think, but they write—not because they understand, but because understanding is not the source of writing, but its effect.

What emerges from this convergence is a vision of authorship no longer tied to subjectivity but to system. To write is to enter a symbolic chain, to add a node to a network that is already in motion. Human or artificial, each act of composition contributes to the evolving architecture of meaning.

Conclusion: From Authorship to Symbolic Continuum

In revisiting Barthes through the lens of intelligent software, we find not the death of writing, but its emancipation from the myth of origin. The author has never truly stood alone. What AI reveals is not a rupture in the history of language but its continuation. As Barthes, Derrida, Peirce and Saussure help us see, meaning arises from within the text, from its differences, its iterations, its infinite regress of symbols.

Instead of asking whether machines can be authors, we might ask what it means that humans ever thought they could be. In the symbolic economy of language, authorship is a function, not a property. It emerges from circulation, not solitude.

Thus, the rise of AI in writing is not an alien incursion but a mirror held up to language itself. In it, we see what they already discerned: that language lives not in the speaker but in the system; not in intention but in inscription; not in the genius but in the grammar. The author may be dead, but the text is more alive than ever.

Related Post

The Author Is Dead: Barthes' Uncanny Prophecy and the Age of AI

https://derridaforlinguists.blogspot.com/2025/04/blog-post_15.html

References

Barthes, Roland. “The Death of the Author.” In Image-Music-Text, translated by Stephen Heath, 142–148. New York: Hill and Wang, 1977.

Foucault, Michel. What Is an Author? In Language, Counter-Memory, Practice, edited by Donald F. Bouchard, translated by Donald F. Bouchard and Sherry Simon, 113–138. Ithaca: Cornell University Press, 1977.

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

McGann, Jerome. The Textual Condition. Princeton: Princeton University Press, 1991.