The Author-Function

In the [YOU] on AI Field Guide

The cycle’s treatment of the author-function focuses on its most profound disruption: the dissolution of what Foucault called the metaphysics of origin. The author-function does not merely organize texts; it originates them—designating a point of departure, a consciousness from which the text emerged, a source that explains the text’s existence. When a text emerges from the collision between human intention and machine inference, the origin becomes distributed. The text is not authored in the sense the metaphysics of origin requires; it is produced through a process with no single point of departure, no sovereign consciousness from which every sentence emerged, no identifiable moment of creation attributable to one agent rather than another.

The cycle draws its practical consequence from the fourth and most transparent of the four functions: the commercial function reveals with particular clarity that the author-function has never been primarily about the relationship between a text and a consciousness. It has been about institutional needs, and those needs persist regardless of how the text was produced. This clarity is, paradoxically, the strongest available argument for maintaining the authenticating function in a new form: since the commercial function survives AI co-authorship without difficulty, the question becomes what the authenticating function should now mean. The cycle’s answer, drawn from parrhesia: authentication means commitment rather than origin. The human author authenticates by standing behind the text’s claims, accepting its consequences, binding herself to the truth of what she has helped bring into the world. The machine can produce statements; it cannot commit to them.

Michel Foucault

Origin

Foucault delivered “What Is an Author?” as a lecture at the Collège de France in 1969, responding to Roland Barthes’s provocative essay “The Death of the Author” (1967). Where Barthes had announced the author’s death as a liberation of the reader, Foucault reframed the question genealogically: not whether the author exists but what institutional work the author’s existence performs. The shift from ontology to function is characteristic of his method and decisive for its application: the question is not whether machines can be authors in some philosophical sense but which institutional functions the author-function currently performs, which of these AI disrupts, and what arrangements will replace those it displaces.

The concept has roots in two strands of Foucault’s broader project. From his archaeological work, it inherits the insight that the conditions of possibility for knowledge are institutional rather than natural—that the category of the author, like the category of madness or the category of the patient, is a historical construction that does work within and for specific institutions, not a pre-institutional datum that those institutions merely recognize. From his genealogical work, it inherits the insistence on tracing the power relations that produced the construction and sustain it: the author-function was constructed when texts became property and authors became legally accountable for content, and its reconstruction in the AI era will be determined by the power relations governing who designs the new arrangements and in whose interest.

Key Ideas

Four functions, four disruptions. The classificatory function is strained because the coherent oeuvre organized around one consciousness no longer holds when contributions are distributed between human intention and machine inference. The legal function reveals itself as a political decision rather than a philosophical derivation—someone must be liable, and who is determined not by analysis but by institutional power. The authenticating function loses its metaphysical grounding but gains a new practical meaning in the form of commitment. The commercial function is the least disrupted and the most revealing: the market still needs names, still needs brands, and the function’s persistence makes visible that it never depended on the authorial consciousness it appeared to require.

The distributed origin. AI collaboration dissolves the metaphysics of singular origin more thoroughly than any previous challenge to authorship. The text that emerges from human-machine collaboration has no single point of departure. Its origin is the collision itself—the space between human intention and machine inference—and the author-function, constructed on the assumption of singular origin, cannot accommodate the distribution without visible strain. The strain is not a failure; it is the visibility of contingency that naturalization had concealed.

Commitment as the residual human function. If origin can be distributed, authentication cannot be abandoned without consequence: without it, no text can be trusted, and the entire apparatus of knowledge-making that depends on texts breaks down. The cycle’s proposal, following Foucault’s analysis of parrhesia, is that authentication migrates from origin to commitment: the human co-author authenticates by standing behind the text’s claims with her career, her reputation, and her willingness to be held responsible for its truth. This is a smaller claim than the Romantic author’s sovereignty but a real one—and one that the machine structurally cannot make.

Roland Barthes