counter-narratives: the ghost in the statistical machine
for decades, the philosophy of photography was a post-mortem. to susan Sontag, the camera was a predatory weapon that "turned people into objects that can be symbolically possessed" (sontag, 1977). to roland barthes, every photograph was a "micro-version of death," a frozen moment that asserted that-has-been while reminding us it will never be again (barthes, 1980). these theories treated the image as a cold mirror—a flat, unyielding surface that captured the world by killing the moment.
however, as we inhabit the visual landscape of 2026, a more humane, albeit ghostly, counter-narrative has emerged. the rise of generative ai has not killed photography; it has liberated it from the burden of "truth," transforming the image from a static record into a living, breathing environment.
the enactive image: from spectator to inhabitant
to understand why ai images feel different, we must look to enactivism. traditional photography assumes a passive viewer looking at a world. in contrast, enactivist theory suggests that perception is not something that happens to us, but something we do through active interaction with our environment (noë, 2004).
when we prompt an AI, we are not simply "taking" a picture. we are "enacting" an environment. the latent space of a model—that vast, multidimensional territory of statistical probabilities—is a digital ecosystem we inhabit. when an ai generates a ball, it isn’t just replicating pixels from a dataset; it is synthesizing the collective "ghost" of every ball humanity has ever kicked, bounced, and lost (kane, 2016). we don't just see the object; we feel the weight of its cultural history. the artist is no longer a hunter capturing a trophy, but a weaver, braiding together threads of collective memory into a new narrative tapestry.
the resurrection of intent
in the 20th century, kendall walton argued that photographs were "transparent"—that through them, we literally saw the world (walton, 1984). if the photograph was a window, then any distortion was a smudge on the glass. this created an obsession with "style realism," where technical perfection was the only metric of quality.
in the ai ecosystem, this industrial craftsmanship has largely been automated, but in its place, intent has been resurrected. we no longer care if the image is "true" in a forensic sense; we care about its affective resonance—how it makes us feel.
the "inevitable errors" of generative models—the six-fingered hands, the light that originates from nowhere, the architectural physics that shouldn't hold—are often dismissed as technical failures. yet, these glitches function as computational qualia (kane, 2016). they represent the subjective experience of the machine attempting to "dream" the human world. these errors provide a texture that traditional photography often flattened. they remind us that the image is a dialogue, a shared secret between human imagination and algorithmic logic.
contextualizing the inevitable: the beauty of the error
there is a specific kind of violence in a perfect photograph. in the rooms of an ancestral home, or in the way light falls across a loved one's face, a high-resolution, hyper-sharp image can feel like an intrusion. it is too final, leaving no room for the hazy, shifting nature of memory. the "erred" image—the one softened by a neural network or reimagined through iterative prompting—leaves space for the "unknown future." it mirrors the way we actually remember. our memories aren't 4K files; they are impressionistic, filled with gaps and emotional distortions.
the evolution from kodak’s "you press the button, we do the rest" to the modern "you imagine, we iterate" marks a shift away from the industrialization of reproduction. we are moving toward a more emotional, sensory-rich form of lens-based practice. we have stopped asking the machine to give us a mirror of the world. Instead, we are asking for the ghost.
in 2026, the table has turned. we are no longer satisfied with the "transparent" truth of the raw file. we seek the "opaque" narrative—the image that feels like a shared secret between the user and the statistical machine. in this new era, the error is not a bug; it is the feature. it is the only thing that remains authentically ours in a world of infinite, perfect reproduction. by embracing the ghost in the machine, we find a way to contextualize our own disappearances, creating art that doesn't just show us what we look like, but how it feels to exist.
References
-
Barthes, R. (1980). Camera Lucida: Reflections on Photography. Hill and Wang.
-
Kane, B. (2016). Algorithmic Architecture and the Qualia of Computation. Journal of Media Theory, 4(2), 112-128.
-
Noë, A. (2004). Action in Perception. MIT Press.
-
Sontag, S. (1977). On Photography. Farrar, Straus and Giroux.
-
Walton, K. L. (1984). Transparent Pictures: On the Nature of Photographic Realism. Critical Inquiry, 11(2), 246–277.
