Studium, Punctum, and AI’s Imaginary: Barthes in the Age of the Machine

The nun. AI generated image

Introduction

In 1980, Roland Barthes published Camera Lucida, a book that would become one of the most influential reflections on photography in the 20th century. There he formulated two key categories—studium and punctum—that aimed to explain the aesthetic and affective experience of the image. More than forty years later, the visual horizon has radically changed: traditional photographs coexist with images generated by artificial intelligence, where there is no lens, no camera, and no “reality” in the classic sense. This raises an unavoidable question: can studium and punctum be carried over into the universe of algorithmic creation? Or must they be redefined to account for a new aesthetic sting in a digital world?

Barthes and Photography: Studium and Punctum

Barthes distinguished between two modes of relating to photography. The studium belongs to the cultural and shared field: “Studium is an application to a thing, a taste for someone, a kind of general, enthusiastic commitment, but without special acuity” (Barthes, 1980, p. 48). It includes what the photographer intended to convey, the historical frame, the recognizable artistic codes.

The punctum, on the other hand, is what escapes intention and emerges as an unexpected detail that pierces the viewer: “The punctum of a photo is that accident which pricks me (but also bruises me, is poignant to me)” (Barthes, 1980, p. 49). Not every image has it; when it does, it acts like a needle piercing perception and memory.

A frequently cited example is the photograph of a group of Nicaraguan soldiers where, by chance, a nun appears in the background. That accidental detail, unforeseen by the photographer, constitutes the punctum: a point that wounds because it was real, because it was there. In fact, the force of the punctum lies in the ontological dimension of photography, its tie to the past and to what Barthes summarized with the formula “ça a été” (“that has been”).

From Eye to Algorithm: Studium in the Age of AI

The scenario changes when reality is no longer captured with a camera, but images are generated through algorithms. Here, studium can be identified with the intention of the user writing a prompt. Just as the photographer once chose frame, subject, and technique, today the prompt engineer defines textual descriptors that guide the model’s work.

Moreover, studium extends to the cultural archive contained in the datasets on which AI was trained. There we find accumulated styles, conventions, visual genres, and recognizable aesthetics that condition the results. The viewer can read these elements and situate them in a cultural horizon, much as they would with photography.

In this sense, studium remains what is legible and comprehensible, though now mediated not by the photographer’s eye but by technical writing and the machine’s vast statistical memory. Here echoes Vilém Flusser’s notion of the “program” of the photographic apparatus: a set of pre-coded possibilities. AI radicalizes this principle by exponentially expanding the range of variations.

The Unexpected in the Machine: Algorithmic Punctum

The most delicate issue is the status of punctum in the digital age. In photography, it was the chance detail of the real world. In AI, there is no such link to what “has been.” Yet a margin of the unexpected remains, now produced by the opacity of the algorithm.

The algorithmic punctum may take several forms:

·         An emergent detail: a strange gesture in a face, an unforeseen shadow, a color out of place.

·         A disturbing error: the infamous deformed hands or distorted faces that appear in some outputs.

·         The mystery of the technical process itself: the sheer incomprehensibility of how AI translates text into a convincing image.

Rather than “real” chance, this is statistical emergence. And yet, for the viewer, the affective experience can be similar: the aesthetic sting persists, though it arises elsewhere.

Hallucination as New Punctum

In artificial intelligence, “hallucinations” occur when the model produces results that do not match the input or lack verifiable grounding. In images, this might be an impossible object, incoherent architecture, or an attribute contradicting the prompt.

At first glance, this seems far from Barthes’s punctum: it is not a real-world detail that surprises, but a misfired correlation. Yet when hallucination provokes an affective reaction in the viewer—unease, strangeness, fascination—it can play that piercing role.

The difference is crucial: photography declared “that has been.” The AI image may instead suggest the opposite: “this has never been.” And still, that “never” is not without effect; on the contrary, it wounds because it shows the impossible or the monstrous. Perhaps it is best to say that it does not deny the referent but shifts it into the realm of simulation: this could have been. Here the reflection approaches Baudrillard, for whom the digital image does not represent but becomes an autonomous simulacrum.

Conclusion

The concepts of studium and punctum shed light on the aesthetic experience of AI-generated images, though with significant displacements. The studium manifests in the prompt and in the cultural archive that organizes production. The punctum is transformed into emergent detail, algorithmic error, or even hallucination.

The sting is no longer the guarantee that “that has been,” but an opening onto an uncertain territory where the opacity of technique replaces the chance of the world. In the passage from photography to AI, the referent becomes decentered: the wound no longer comes from real pastness but from digital enigma.

The final question is inevitable: do we need a new term, beyond punctum, to name this artificial sting that wounds us even though “it has never been”?

References

·         Barthes, R. (1980). La chambre claire. Note sur la photographie. Paris: Gallimard/Seuil.

·         Barthes, R. (1989). Camera Lucida. Trans. Joaquim Sala-Sanahuja. Barcelona: Paidós.

·         Batchen, G. (1997). Burning with Desire: The Conception of Photography. MIT Press.

·         Flusser, V. (2002). Towards a Philosophy of Photography. Madrid: Síntesis.

·         Steyerl, H. (2009). In Defense of the Poor Image. e-flux Journal, 10.

·         Baudrillard, J. (1981). Simulacres et simulation. Paris: Galilée.