The Artist’s Body

Barthes declared the author dead in 1967, and generative AI has made the diagnosis literal. His argument was that a text is not the expression of a singular consciousness but a tissue of quotations drawn from innumerable centres of culture, and that meaning arises not from what the author intended but from what the reader makes of the encounter. Large language models enact this theory with an almost embarrassing directness: they assemble text from patterns absorbed across vast corpora, without intention, without biography, without a body. If we accept Barthes's framework — and the Dadaists gestured toward it half a century earlier, when Tristan Tzara proposed making a poem by cutting words from a newspaper and drawing them from a hat — then an AI system can be an author in at least the functional sense. It curates a mechanism that produces meaning, and meaning, as Barthes insisted, happens at the destination, not the origin. The reader completes the work regardless of who or what initiated it.

This line of argument is coherent, and it clears space for something important: the recognition that authorship, understood as the arrangement of cultural materials into a form that produces meaning in a receiver, is not an exclusively human activity. Duchamp demonstrated this with the readymade — a urinal placed in a gallery becomes art not because of what was made but because of the decision to frame it, and because of what happens in the viewer's mind when they encounter the framing. AI can perform this operation. It can select, arrange, and present. It can, with increasingly fine calibration, curate the mechanism. And in most professional contexts — brand work, content production, strategic communication — this is sufficient. The question of whether the arrangement was made by a person or a probabilistic model is, for practical purposes, immaterial. The work works or it does not.

But Barthes killed the author precisely to liberate the text. He was making a claim about interpretation, not about cultural production as a whole. And when we move from the author — the function that produces the text — to the artist — the figure that inhabits public life — the Barthesian framework runs out of explanatory power, because the artist is not reducible to the work they produce. The artist is a biography.

AI can be an artist, but not this kind of artist

Consider what Lady Gaga means to her audience. The music can be analysed semiotically; the visual language can be mapped; the political positions can be catalogued. All of these could, in principle, be emulated by a sufficiently capable AI system. A hypothetical AI celebrity could produce songs in her idiom, adopt a visual grammar of similar complexity, and articulate progressive political stances with comparable clarity. The parasocial infrastructure — the fan accounts, the stan culture, the identification with the persona — could even be sustained for a time, because research on parasocial relationships shows that audiences form emotional bonds with fictional characters and animated entities as well as with living people (Horton and Wohl 1956; Liebers and Schramm 2019). There is nothing in the structure of parasocial attachment that strictly requires a biological human at the other end.

And yet something would be missing, and the missing thing is not a mystical quality or a vague appeal to the soul. It is mortality. What makes Billie Eilish's openness about anxiety and depression resonate is not the content of her disclosures — a language model could produce equivalent disclosures with persuasive emotional texture — but the knowledge that these disclosures are anchored in a life that is finite, contingent, and at risk. When she speaks about vulnerability, the audience understands that this vulnerability is real in a specific sense: it is the vulnerability of a body that can be hurt, a career that can end, a person who will age and eventually die. The fan's relationship to the artist is shaped at a deep level by this awareness, even when it is not consciously articulated. The aspiration the fan feels is not simply toward the artist's aesthetic — it is toward the way a mortal person navigates difficulty, takes public positions at personal cost, and produces work that is marked by the passage of time.

This is where the figure of the artist diverges from the function of the author. The author, in Barthes's sense, is a textual effect — a position within a system of signs. The artist is a temporal being whose work acquires meaning partly because it is produced within the constraints of a life. David Bowie's late album Blackstar, released two days before his death, is not simply a collection of songs; it is an act of reckoning with mortality that acquires its force from the listener's knowledge that its creator was dying as he made it. No AI system can occupy this position, because no AI system has a life to lose. The work an AI produces may be sophisticated, moving, even formally original — but it cannot be a testament, because a testament requires a witness who is also a participant, someone who speaks from within the condition they describe.

Eilish’s TikTok behind the scenes: not the song, nor the image, nor the style: the art is the artist; the embodiment of its humanity, transcending the art object.

Eilish’s TikTok behind the scenes: not the song, nor the image, nor the style: the art is the artist; the embodiment of its humanity, transcending the art object.

All the way to Dada

The Dadaists understood this distinction even as they challenged the conventions of authorship. Tzara's newspaper-poem technique was deliberately anti-authorial: it removed the artist's hand from the selection of words and surrendered the outcome to chance. But Tzara himself (performing the technique, publishing the results, arguing for its significance in manifestos and public provocations) remained a figure. The method was mechanical; the stance was biographical. The Dada movement's power derived not from the works it produced, which were often deliberately ephemeral, but from the fact that living people chose to make those works, in a particular historical moment, as an act of refusal that carried personal consequences. The curation of the mechanism was provocative precisely because a mortal person chose to curate it.

The inseparable body

AI can produce the mechanism. It cannot yet produce the figure. And the figure matters because art is not only a semiotic event — a message decoded by a receiver — but also a social relationship between a person who has made something and a person who encounters it, mediated by the awareness that both of them are finite. This is what the parasocial research circles around without quite naming: the reason fans grieve when artists die is not that the supply of new work has ended, but that the relationship, which was structured by the shared condition of mortality, has been severed on one side. When Robin Williams died, the intensity of public mourning reflected not just the loss of future comedies but the collapse of a parasocial bond that was built on the audience's sense that his humour was a response to his suffering; suffering that was real because it was happening to a body, in a life, under the pressure of time.

The originality question

For the purposes of this newsletter's ongoing argument about originality, the distinction matters in a specific way. Originality of ideas, as the previous essays have argued, is the scarce and compounding input in an AI-augmented creative economy. But the figure of the artist adds a second dimension: originality of position. The artist occupies a position that is defined not only by what they make but by who they are while making it — by the biographical fact that they are a particular person, in a particular moment, bearing particular risks. This positional originality is something that AI cannot replicate, because it requires the one thing that machines do not have: stakes. A human artist who takes a controversial public stance is doing something that an AI cannot do, because the human can lose something — reputation, safety, relationships — while the AI cannot. The audience knows this, and this knowledge shapes the meaning of the work.

Barthes was right that the text does not need an author to produce meaning. But culture, understood as a living system of relationships between people who make things and people who receive them, has always needed something more than texts. It has needed figures — people whose work is legible partly because their lives are visible, and whose lives are significant partly because they are temporary. AI has entered the space of the author. It has not entered the space of the artist. And the difference between those two spaces may turn out to be the most durable boundary in the entire conversation about what remains human in creative work.

---

Sources to cite:

• Barthes, R. (1967/1977), "The Death of the Author," in Image, Music, Text, trans. Stephen Heath, Fontana.
• Duchamp, M. (1917), Fountain — the readymade as authorial gesture.
• Tzara, T. (1920), "To Make a Dadaist Poem," from Dada Manifesto on Feeble Love and Bitter Love.
• Eco, U. (1962/1989), Opera Aperta / The Open Work, Harvard University Press — for the structure of bad taste and the relationship between avant-garde and kitsch (connecting to the previous article).
• Horton, D. and Wohl, R.R. (1956), "Mass Communication and Para-Social Interaction," Psychiatry, 19(3).
• Liebers, N. and Schramm, H. (2019), "Parasocial Interactions and Relationships with Media Characters," Communication Research Trends, 38(2).
• Craig, C. and Kerr, I. (2021), "The Death of the AI Author," Ottawa Law Review, 52(1) — for the ontological argument against AI authorship.
• Foucault, M. (1969/1977), "What Is an Author?" — counterpoint to Barthes, arguing that the author-function still organises how texts are received and regulated.