Autographic and Allographic Arts

In the AI Story

Goodman developed the distinction in response to the forgery problem: why does it matter, aesthetically, that a painting is a forgery if the forgery is perceptually indistinguishable from the original? His answer: in autographic arts, the history of production is not merely of biographical interest—it is constitutive of what the work achieves in its art-historical context. Knowing that van Meegeren, not Vermeer, painted The Supper at Emmaus changes what the painting does: it is no longer a seventeenth-century breakthrough but a twentieth-century pastiche. The painting looks the same, but what it exemplifies, what it contributes to the development of pictorial representation, changes when its history changes. In allographic arts, no comparable change occurs. A performance of Beethoven's Ninth is evaluated by the same criteria whether it is the Berlin Philharmonic or a high school orchestra, because the work's identity is determined by the score, not by who performs it.

The classification seemed stable for three centuries of print culture. Literature was safely allographic: the text is the text, and any correct copy is an instance of the same work. But the classification becomes unstable in the age of AI, where the productive history of texts is no longer transparent. When a text is produced through human-AI collaboration—when some sentences originate with the human and others with the machine, and the boundary is invisible even to the participants—the allographic criterion (identity determined by the text) conflicts with the reader's expectation that the name on the cover identifies the person who produced the sentences through the cognitive labor of writing. The text is the same. The authorial structure is different. And if authorship matters to what the text achieves, then literature is experiencing an autographic crisis: the sudden relevance of productive history in a domain where productive history was previously irrelevant.

Goodman's framework does not resolve the crisis—it was designed for a world in which productive history was either transparently single-authored or explicitly collaborative (as in musical performance). But the framework identifies what is at stake with unusual precision. The question is not whether AI-generated texts can be good literature—Goodman's functionalism permits that they can. The question is whether texts with indeterminate authorial structures—where the human cannot specify which aspects of the worldmaking originated with her and which with the machine—belong to a new category requiring new evaluative standards. The autographic/allographic distinction was designed for a world with two clear categories. The AI age has produced a third: the indeterminately authored work, which has the surface properties of allographic art (the text determines identity) and the evaluative structure of autographic art (origin matters to what the work achieves).

Origin

The distinction was introduced in Languages of Art (1968), Chapter III, titled 'Art and Authenticity.' Goodman argued that the aesthetic relevance of authenticity—the question of whether a work is genuine or forged—depends on the medium. In painting, authenticity is aesthetically decisive; in music and literature, it is not. The difference is not sentimental but structural: painting lacks a notation that could specify the work's identity independently of its productive history, while music and literature possess such notations (the score, the text). The argument was buttressed by the formal analysis of what notation requires—syntactic and semantic differentiation, unambiguity of compliance—and by the demonstration that painting and sculpture lack these properties while music and literature (mostly) possess them.

Key Ideas

History of production is constitutive in autographic arts. A forged Rembrandt is not a Rembrandt, even if perceptually identical, because the work's identity includes its productive history—who made it matters to what it is.

Notation determines allographic identity. In allographic arts, the work's identity is fixed by compliance with a notational specification (score, text)—any correct instance is a genuine instance of the same work.

Literature's autographic crisis. The allographic classification of literature assumed transparent productive history; AI makes productive history opaque, forcing the question of whether origin matters to literary evaluation.

A third category emerges. AI-collaborative works are neither cleanly autographic nor allographic—indeterminate authorial structures resist the binary classification Goodman's framework was built to maintain.