Ceci n’est pas une article.
Authorship in AI-generated texts represents the logical culmination of the conceptual evolution that began with Duchamp and was theorized by Benjamin, raising radically new questions about the very nature of creation.
AI as Textual Ready-made
Following Duchampian logic, when someone uses AI to generate text, they are performing a conceptual gesture similar to the ready-made: they select, contextualize, and present an output generated by a system that processes and recombines millions of pre-existing texts. The “prompt” becomes the equivalent of the “R. Mutt” signature: the gesture that transforms processed data, premade objects, into content with authorial or artistic intention.
Extreme Reproducibility
Benjamin theorized about the loss of “aura” in mechanical reproduction, but AI takes this to the extreme. Reproduction through mechanical procedures does not strictly produce copies but reproductions: AI does not copy existing texts, but rather reproduces linguistic patterns learned from vast textual resources. Each output is simultaneously unique (never repeated exactly) and multiple (generated from the same base models).
The Death of the Digital Author
Roland Barthes’ theory of the “death of the author” acquires new relevance: in AI-generated texts, who is really the author?
- The prompt user? Acts as a Duchampian “selector”.
- The model programmers? Create the “reproductive machine”.
- The training corpus authors? Their “ghost” texts permeate the output.
- The AI itself? Can it be considered a creative agent?
What role does the intention play in the creative process? Shouldn’t it be the most important thing?
Author as Operator
In photography and filming, the author becomes someone who operates reproductive machines. Each performance is unrepeatable, while in cinema the performance can be repeated as many times as necessary. It can be tested, assembled from different takes, and then exhibited infinitely.
Text as “Multidimensional Space”
Barthes’ prediction materializes literally: the text presents itself as a fabric of citations and references to countless centers of culture. AI models literally weave texts from patterns extracted from millions of sources, creating that “multidimensional space” that Barthes described theoretically.
New Categories of Authorship
AI generates hybrid categories that transcend traditional distinctions:
Asymmetric Collaborative Authorship: The human contributes intention and context, AI contributes generative capacity. Neither can create the final result without the other.
Emergent Authorship: The final text did not exist in the user’s mind nor in the training data: it emerges from the interaction between both.
Distributive Authorship: It extends across multiple agents: user, developers, corpus authors, data curators, human reviewers who refine the models or the final result.
Contemporary Philosophical Context
Just as the interwar crisis challenged the foundations of modernity, the digital revolution raises fundamental questions regarding the essence of human creativity:
- Is originality an illusion if all text arises from pre-existing patterns?
- Does the nature of authorship change if the creative process is automated?
- What does it mean to be “creative” in the age of artificial intelligence?
- And who can claim for that creativity?
Practical and Ethical Implications
This evolution generates tensions similar to those faced by the avant-garde:
- Legal: Legal systems struggle to define intellectual property of AI-generated content.
- Economic: How is “authorship” compensated when the process involves multiple agents?
- Cultural: What value does human expression have when machines can imitate it?
The Final Democratization
AI represents the culmination of the artistic democratization initiated by Duchamp: anyone can “create” sophisticated texts without mastering traditional writing techniques, just as Duchamp demonstrated that anyone could create art without mastering traditional pictorial techniques.
Authorship in the AI era does not eliminate the human author, but transforms them into a conceptual curator: someone who selects, contextualizes, and presents generative outputs with meaningful intention. It is the natural evolution of a trajectory that began when Duchamp signed a urinal and called it art.
In addition, it is impossible to request that an AI “think” about Duchamp’s impact on the concept of authorship if one is unaware of his existence.