1826
Invention, announcement, and the first theoretical claims
A note on retroactive history: none of the writing in this block, or the one that follows, was produced as "photography theory." The people writing were making aesthetic arguments, defending commercial interests, practicing science, or producing journalism. The field of photography theory as an academic discipline did not exist until the 1970s. What that decade did, among other things, was reach back into this earlier material and claim it as a theoretical tradition. That retrospective construction is itself a historical act worth noticing. The texts here are genuinely important, but their importance was partly assigned after the fact.
1826–27
Photography / Invention
Niépce, View from the Window at Le Gras
Joseph Nicéphore Niépce produced the earliest surviving photograph around 1826 or 1827, using a pewter plate coated with bitumen of Judea and an exposure of several hours. The image, showing rooftops and a courtyard in Burgundy, is barely legible to the contemporary eye. Niépce called his process heliography, sun-writing, which already encodes a theoretical claim: light is the author, not the operator. He later entered a partnership with Daguerre, who refined and commercialized the process after Niépce's death in 1833. Photography's public history begins in 1839, but its actual history begins here.
1839
Photography / Philosophy
The announcement and the first ontological claims
When Daguerre's process was announced to the French Academy of Sciences on August 19, 1839, the astronomer Dominique Arago framed it immediately in theoretical terms: the photograph was nature drawing itself, without human intervention. This claim, nature producing its own image, is the founding myth of photographic indexicality. It was a rhetorical strategy as much as a description, designed to secure state funding, but it embedded a theoretical assumption that would take more than a century to fully unpack. The "pencil of nature" framing, which Talbot borrowed for his own title, carries the same weight: the image is caused, not made.
1844
Photography / Philosophy
Talbot, The Pencil of Nature
The first photographically illustrated book, but also the first sustained attempt to think through what photography does. Talbot's notes accompanying the plates are genuinely philosophical: he considers what it means for an image to be produced by light rather than by hand, speculates about photography's future uses (he predicts document copying and surveillance before either existed at scale), and reflects on the uncanny quality of photographic detail. The Pencil of Nature is not a systematic treatise, but it asks questions about automatism, authorship, and the relationship between image and reality that will define the field more than a century later.
1839–1860s
Art Theory / Photography
The art-or-science debate
Almost immediately after the announcement of photography, a debate began about whether it was an art form or a mechanical process, and therefore whether photographers were artists or operators. The stakes were partly commercial (guild protections, copyright, exhibition rights) but also genuinely aesthetic. Charles Baudelaire's essay "The Modern Public and Photography" (1859) is the sharpest early articulation of the anti-photography position: he argued that photography was industry masquerading as art, flattering the laziness of people who wanted likeness without imagination. This debate is never fully resolved; it echoes in every subsequent argument about AI image generation.
1880s–1910s
Photography / Aesthetics
Pictorialism and the defense of photography as fine art
The pictorialist movement produced the first sustained body of theoretical writing explicitly defending photography as art. Peter Henry Emerson's Naturalistic Photography (1889) grounded its argument in Helmholtz's theories of human vision: a photograph should render focus and blur the way the eye actually perceives, not with uniform sharpness. Alfred Stieglitz's journals Camera Notes (1897) and Camera Work (1903–17) published theoretical essays and criticism that collectively made the case for photography's artistic seriousness. These are real arguments about vision, aesthetics, and medium specificity, made from within photographic practice rather than the academy.
1870s–1900s
Photography / Power / Classification
Galton, Bertillon, and the archive of bodies
Francis Galton's composite photography and Alphonse Bertillon's criminal identification system both embedded theories of what photographs could reveal about social categories, criminality, and race. These are not aesthetic arguments but epistemological ones: the photograph as instrument of knowledge, classification, and social control. Allan Sekula's "The Body and the Archive" (1986) is the essential retrospective account, reading this material through Foucault's framework of power-knowledge. The surveillance and classification functions of early photography are now understood as foundational to how the medium operates institutionally, and as direct precursors to the biases built into visual datasets and facial recognition systems.
The animating question of this era
"What is this new thing? Is it art, science, or industry? Does the camera reveal truth or construct it? And what can a mechanical image do that a hand-made image cannot?"
Questions to consider
The founding claim of photography, that nature draws itself without human intervention, is already a theoretical position about indexicality. How much of what later theorists debated was already implicit in the way photography announced itself in 1839? And how does the art-or-science debate of the 1840s map onto the art-or-technology debate about AI image generation today?
1920s
Documentary, film theory, and the photograph as historical object
The interwar period produced some of the sharpest early theoretical writing on photography, though again not from within a photography field. Benjamin was writing cultural criticism and political theory. Kracauer was writing journalism and sociology. The documentary photography movement was driven by political urgency, not academic theorizing. Film theory was developing in parallel, asking many of the same questions about mechanically produced images. All of this material was available to the 1970s field, which absorbed it selectively, in some cases (Benjamin) very heavily, in others (Kracauer) considerably less so.
1931
Photography Theory / Cultural Criticism
Benjamin, "A Little History of Photography"
A shorter and in some ways more useful text than the "Work of Art" essay for thinking about photography specifically. Benjamin introduces the optical unconscious: photography reveals aspects of the visual world that the human eye cannot consciously perceive, in the way that psychoanalysis reveals aspects of mental life that consciousness cannot access. He also writes about aura in the context of early portrait photography, where long exposure times required a kind of presence and duration that later fast photography eliminated. The essay is attentive to photography's history in a way the "Work of Art" essay is not, and it asks what is lost as well as what is gained when the medium accelerates.
1927
Photography / Sociology
Kracauer, "Photography"
Published in the Frankfurter Zeitung, this is one of the sharpest early critiques of photographic meaning and one of the most underread texts in the canon. Kracauer argues that the photograph captures spatial configuration but not meaning: it records the surface of things without their historical significance. A photograph of a grandmother as a young woman does not convey who she was; only memory, which is selective and saturated with meaning, does that. His argument anticipates Barthes's Camera Lucida by more than fifty years, and his distinction between the spatial organization of the photograph and the temporal organization of memory is directly relevant to any account of AI-generated images, which have spatial configuration but no history at all.
1935–44
Photography / Politics
The FSA project and the documentary debate
The Farm Security Administration photography project, directed by Roy Stryker and including Dorothea Lange, Walker Evans, and Gordon Parks, raised urgent questions about the relationship between documentary photography and political action. Could a photograph of poverty produce political change, or did it aestheticize suffering in a way that neutralized it? Walker Evans's austere, formally rigorous images implied one answer; Lange's emotionally direct images implied another. James Agee and Walker Evans's Let Us Now Praise Famous Men (1941) is partly a prolonged meditation on the ethics of documentary representation. These debates were not academic, but they were theoretical in their stakes, and they fed directly into photo theory's later discussions of documentary ethics and the politics of representation.
1920s–30s
Film Theory / Photography
Early film theory: Balazs, Arnheim, Kracauer
Film theory developed in the 1920s–30s as a parallel formation, asking many of the same questions about mechanically produced images, realism, and aesthetics. Bela Balazs wrote about the close-up and physiognomy. Rudolf Arnheim's Film as Art (1932) argued that film's formal limitations were the source of its aesthetic possibilities. Kracauer's later Theory of Film (1960) argued that film's proper subject was the redemption of physical reality. These texts are not directly about photography, but they develop a vocabulary for thinking about mechanically produced images and the relationship between medium and meaning that photography theory draws on. The boundary between film theory and photography theory has always been porous.
The animating question of this era
"Can a mechanically produced image tell the truth? What is the relationship between documentary fidelity and political meaning? And what does the camera see that the eye does not, or cannot?"
Questions to consider
Kracauer argued that the photograph captures spatial configuration without historical meaning. If that is true of photographs, what does it mean for AI-generated images, which have spatial configuration but no referent history at all? And the FSA documentary debate, whether images of suffering produce empathy or aestheticize and neutralize it, continues in debates about photojournalism, social media imagery, and AI-generated representations of crisis.
1900s
The foundations that will later be inherited
The disciplines that will eventually constitute photography theory, media studies, and semiotics are forming separately, with no awareness of each other. Saussure lectures in Geneva; the Frankfurt School forms in response to fascism and mass culture; Walter Benjamin begins writing about mechanical reproduction. None of them are writing "about photography" as a field.
1906–11
Semiotics
Saussure's Course in General Linguistics
Saussure never published this himself; it was reconstructed from student notes after his death. His key move: the sign is arbitrary, the relationship between signifier (sound-image) and signified (concept) is conventional, not natural. This will become the theoretical foundation that photography theory later has to argue against, because photographs seem to have a non-arbitrary, motivated relationship to what they depict.
1860s–1910s
Semiotics / Philosophy
Peirce's icon / index / symbol
Charles Sanders Peirce worked in relative obscurity during his lifetime. His tripartite sign theory, icon (resemblance), index (causal trace), symbol (convention), will become central to photography theory in the 1970s–80s. The index is the key term: a photograph is an index because it was caused by light reflected from its subject. This is the origin of all debates about AI images and post-indexicality.
1935
Art Theory / Philosophy
Benjamin, "The Work of Art in the Age of Mechanical Reproduction"
Written in exile from Nazi Germany, Benjamin argues that mechanical reproduction destroys the aura of the original, its here-and-now, its embeddedness in tradition. Photography and film are his primary examples. This essay will be cited continuously for 90 years. Note: Benjamin was not writing as a "photography theorist"; he was writing about fascism, mass culture, and political aesthetics. Later photo theory often extracts him from that context.
1923–
Art Theory / Critical Theory
The Frankfurt School forms
Horkheimer, Adorno, Benjamin, Marcuse. Their central concern: how does mass culture under capitalism produce consent and foreclose critical thinking? The culture industry concept (Adorno and Horkheimer, 1944) will permanently shape how media studies and art theory think about images, reproduction, and ideology. Photography is always implicated; it is the mass culture image par excellence.
The animating question of this era
"What is the relationship between the sign and the real? Is meaning natural or constructed? And what happens to art when it can be mechanically reproduced infinitely?"
Questions to consider
If the photograph’s meaning depends on the structural difference between types of signs, icon, index, symbol, what happens when a single image combines all three? When you look at an AI-generated image that resembles a photograph, which type of sign are you encountering? Does it matter?
1950s
Structuralism, mass media, and the first accounts of the photograph
French structuralism transforms the humanities. Barthes begins writing about photography and mass culture. McLuhan inaugurates media theory in North America. The first sustained theoretical accounts of the photograph as something other than a document begin to appear, but they come from literary critics and cultural commentators, not from a "photography field" that doesn't yet exist.
1957
Semiotics / Cultural Theory
Barthes, Mythologies
Barthes analyses mass cultural objects, wrestling, steak and chips, the face of Greta Garbo, using a Saussurean framework. His concept of myth: a second-order sign system that naturalizes ideology, making historical constructions appear inevitable and natural. This is one of the first sustained attempts to apply semiotic method to visual culture. Photography appears throughout as an example of how myth works.
1961 / 1964
Semiotics → Photo Theory
Barthes, "The Photographic Message" and "Rhetoric of the Image"
Two essays that will become foundational for photo theory. "The Photographic Message" introduces the idea of the photograph as a "message without a code", a paradox, because the image appears purely denotative (just showing what's there) while actually operating at the level of connotation. "Rhetoric of the Image" introduces the concept of text's anchorage, the idea that caption and image constrain each other's meaning. This is the Barthes that contemporary AI-image practice most directly unsettles.
1962 / 1964
Media Theory
McLuhan, The Gutenberg Galaxy and Understanding Media
"The medium is the message." McLuhan argues that the form of a medium; not its content, is what shapes human perception and social organization. This is scandalous to critics on the left (it seems to evacuate politics and ideology from media analysis) but enormously generative for media theory. It establishes the idea that technical apparatus matters, an idea that will later connect to Flusser's apparatus theory and, eventually, to AI as apparatus.
1960–68
Art Theory / Practice
Conceptual Art and the dematerialization of the object
Artists like Sol LeWitt, Lawrence Weiner, Joseph Kosuth begin producing work where the idea is the art, not the object. Lucy Lippard and John Chandler coin "dematerialization" in 1968. Photography becomes central to Conceptual Art not as an art form in itself but as documentation, a neutral carrier of the concept. This will create a lasting tension in photo theory: is the photograph a document or an art object? A question that AI image generation reopens again.
The animating question of this era
"How does the image produce meaning? Is the photograph different from other signs because it looks like what it depicts? And is the medium itself, independent of content, what actually structures experience?"
Questions to consider
Barthes argued that text anchors the meaning of an image; it tells you how to read what you see. But what happens when language generates the image rather than describing it afterward? Does the relationship between text and image change fundamentally, or is it still a form of anchorage?
1960s
Post-structuralism: the critique from within
A generation of French thinkers, working in the wake of structuralism and in the political ferment of 1968, begin dismantling structuralism's own assumptions. If structuralism said meaning is produced by stable systems of difference, post-structuralism says those systems are never stable: internally contradictory, historically contingent, saturated with power. Derrida, Foucault, and Lacan each attack a different pillar. Phenomenology offers a parallel alternative; not systems but lived, embodied experience.
1966 / 1967
Post-Structuralism
Derrida: différance, the trace, and deconstruction
Derrida's Of Grammatology (1967) attacks the foundation of structuralism: meaning is never anchored by a stable center. His concept of différance (combining "to differ" and "to defer") argues that meaning is always deferred along an endless chain of differences, never fully present. The trace is what remains when presence is always-already absent. For photography theory this creates a precise problem: the photograph's trace of the real is not the simple presence Barthes' ça-a-été implies but is itself caught in deferral. Derrida's trace ≠ Peirce's index, a distinction that matters for any account of AI image generation.
1966 / 1969
Post-Structuralism / Power
Foucault: discourse, power-knowledge, and the archive
Foucault's The Order of Things (1966) shifts attention from what is said to the rules determining what can be said, the episteme, the underlying order of knowledge in any given era. His concept of discourse: the entire system of practices and institutions that produces objects of knowledge. And the archive: not a building of documents but the system of rules governing what is preserved and sayable. Batchen's genealogy of photography is explicitly Foucauldian. Crawford and Paglen's critique of training sets is implicitly Foucauldian, a dataset is an archive in Foucault's sense, a power-knowledge system determining what images can be generated.
1949 / 1960s–70s
Post-Structuralism / Psychoanalysis
Lacan: the gaze, the mirror stage, the split subject
Lacan reread Freud through structuralist linguistics: the unconscious is structured like a language. His mirror stage: the infant first recognizes itself in a mirror image, and this misrecognition founds the ego. The image precedes and constitutes the self. His concept of the gaze: not the look of a subject at an object, but the point from which the subject feels itself seen, an object in the visual field. The gaze looks back. Mulvey will deploy this for feminist film theory (1975). The Lacanian gaze is essential for understanding the politics of AI-generated imagery of bodies, whose bodies appear in training data and in what contexts.
1967 / 1969
Post-Structuralism / Authorship
Barthes, "Death of the Author" / Foucault, "What is an Author?"
Two essays that permanently changed how theory thinks about authorship. Barthes (1967): meaning is not deposited by the author's intention; it is produced by the reader. The biographical author is irrelevant; what matters is the scriptor, a function of language. Foucault (1969): the "author function" is a legal and institutional category governing circulation and ownership of discourse. Both are directly relevant to AI image generation: who is the author of a prompt-generated image? The prompter, the model, the training set contributors, the engineers? Artist practice can perform this question rather than resolve it.
1945 / 1960
Phenomenology
Merleau-Ponty: embodied perception and "Eye and Mind"
Merleau-Ponty's Phenomenology of Perception (1945) argues that perception is irreducibly embodied, always the perception of a body in a world, not a disembodied mind processing inputs. The body is the medium through which we have a world at all. His essay "Eye and Mind" (1960) applies this to painting: the painter doesn't represent the visible world, they make visible what couldn't otherwise be seen. Phenomenology offers an alternative to semiotic photo theory: instead of asking how photographs produce meaning through codes, it asks what it is like to encounter a photograph as a lived body. Barthes' Camera Lucida is partly phenomenological: the punctum is a bodily, pre-semiotic response.
1949 / 1963–70
Feminist Theory (Foundations)
De Beauvoir, second-wave feminism, the construction of gender
Simone de Beauvoir's The Second Sex (1949): "One is not born, but rather becomes, a woman." Femininity is socially constructed, a direct precursor to Butler's performativity. Second-wave feminism (1960s–70s) creates the political conditions for feminist theory to enter the academy. In art and photography, interventions focus on the representation of women's bodies, exclusion from the canon, and the male gaze as a structuring principle. All of this becomes urgent again with AI image generation trained on datasets in which women's bodies are massively over-represented in particular contexts.
The animating question of the late 1960s
"If meaning is unstable, power structures all knowledge, the author is a legal fiction, and the subject is constituted through misrecognition, what grounds any claim to truth, authenticity, or representation?"
Questions to consider
Derrida argued that the “trace” is never simply present; it always points to an absent origin. How does this complicate photography’s claim to capture a moment that “really happened”? And if an AI image has no origin event at all, does Derrida’s trace concept apply, or does it require something new?
1980s
The "theory boom" and the Pictures Generation
Post-structuralism, Derrida, Lacan, Foucault, floods into art criticism and photo theory. Authorship, originality, and representation become the central critical terms. The Pictures Generation (Sherman, Levine, Prince) makes work that directly enacts these theoretical concerns. Art theory and photography theory become almost indistinguishable. Meanwhile, cultural studies in Britain consolidates a different, more sociological approach to images.
1977–85
Art Theory / Photography
The Pictures Generation
Cindy Sherman, Sherrie Levine, Richard Prince, Barbara Kruger. Artists working with appropriated, staged, and re-photographed images that directly enact post-structuralist theories of authorship and representation. Levine's re-photographs of Walker Evans are the canonical example: what is originality? what is authorship? Douglas Crimp's essay "Pictures" (1977, expanded 1979) theorizes the work. This is art theory and photography theory happening simultaneously through practice, the model of artist-scholar work you operate in.
1981
Photography Theory / Art History
Krauss, "Photography's Discursive Spaces"
Krauss argues that photography was absorbed into art history on art history's terms, which distorted understanding of what photography actually is. Art history reads photographs as aesthetic objects; but photographs were historically embedded in discursive practices (science, geography, portraiture) that gave them meaning. This essay is important for you because it models arguing against how a field has absorbed something, which is structurally similar to arguing against how photo theory has absorbed (or failed to absorb) AI.
1980
Media / Cultural Studies
Stuart Hall, "Encoding/Decoding"
Hall argues that media texts are encoded with preferred meanings by their producers, but decoded by audiences in ways that may align, negotiate, or oppose those preferred meanings. This is a Marxist semiotic framework applied to mass media. It shifts analysis away from the text itself toward the circuit of production, circulation, and reception. For photography theory, it means the image's meaning is never fixed, it's contested at every stage.
1989
Photography Theory
Solomon-Godeau, Photography at the Dock
Essays spanning the 1980s, collected here. Solomon-Godeau brings a feminist and Marxist critique to bear on photography's institutional history, how the medium was legitimized, who was included and excluded, how the "photography world" functioned as an ideological apparatus. Her attention to the politics of the archive and what gets counted as photography anticipates the questions AI training sets now raise.
1975 / 1985
Feminist Theory / Film & Photo
Mulvey's gaze, Haraway's cyborg
Laura Mulvey's "Visual Pleasure and Narrative Cinema" (1975) applies the Lacanian gaze to film: Hollywood cinema structures looking around a male gaze that objectifies women and positions the viewer to identify with a male protagonist. This becomes foundational for feminist photo theory and the analysis of how cameras frame bodies. Donna Haraway's "A Cyborg Manifesto" (1985) is a different kind of feminist intervention: the cyborg as a figure that refuses the boundary between human and machine, nature and culture. Haraway is a direct theoretical ancestor for thinking about human-AI creative collaboration, she made that hybridity politically generative rather than threatening.
1978 / 1983
Postcolonial Theory
Said's Orientalism and the politics of representation
Edward Said's Orientalism (1978) argues that Western representations of "the East" are not neutral descriptions but a system of knowledge-power that constructs the Orient as exotic, inferior, and in need of colonial governance. Photography was a central tool of this construction, colonial photography classified, surveilled, and aestheticized colonized peoples. Homi Bhabha's concept of mimicry (colonial subjects copying colonial culture but never quite fitting) and Gayatri Spivak's question "Can the Subaltern Speak?" extend this to ask: who gets to represent whom, and on whose terms? This is the theoretical framework behind every analysis of racial bias in AI training data.
1990
Feminist / Queer Theory
Butler, Gender Trouble and performativity
Judith Butler's Gender Trouble (1990) argues that gender is not an expression of an inner essence but a performance, constituted through repeated, regulated acts rather than revealed through them. There is no "original" gender that performance copies; the performance is all there is. This concept of performativity migrates into many areas beyond gender: the performativity of documents, of photographs, of identity claims. AI-generated images perform photographic truth without having it; their photorealism is a performance with no original. Butler's framework makes that analysis precise.
The animating question of the 1980s
"If meaning is not intrinsic to the image but constructed through discourse, representation, and ideology, what is left of photography's claim to truth? And who controls the terms of that construction?"
Questions to consider
The Pictures Generation used appropriation and re-photography to expose how images construct rather than reflect reality. How does AI image generation extend, or complicate, that critique? Is prompting a form of appropriation? Who or what is being appropriated?
1990s
The digital turn and the crisis of indexicality
Digital photography arrives and immediately provokes a theoretical crisis: if the photograph can be manipulated pixel by pixel without leaving a trace, what happens to its indexical claim? Visual culture studies emerges as a broader interdisciplinary formation. Postcolonial theory begins to reshape how the field thinks about archives and representation. The internet begins to transform image circulation in ways that theory is only beginning to address.
1990–95
Photography Theory
The digital indexicality debate
William Mitchell's The Reconfigured Eye (1992) argues that digital photography severs the indexical bond, the image can be manipulated without trace, so it cannot function as evidence. This is contested. Fred Ritchin, among others, argues that the social belief in photography's indexicality persists even when the technical basis is undermined. Key insight: the debate reveals that indexicality was never purely technical; it was always also social and institutional. This is what makes multi-dimensional accounts of indexicality viable.
1994–96
Semiotics / Cultural Studies
Visual culture studies emerges
Nicholas Mirzoeff's An Introduction to Visual Culture (1999) and W.J.T. Mitchell's work consolidate "visual culture studies" as a formation broader than art history or photo theory, encompassing film, advertising, digital media, everyday visual experience. It draws on postcolonial theory, feminist theory, and queer theory more explicitly than October-style criticism. The field's founding question: what do images want? (Mitchell's phrase). They are not passive carriers of meaning but active agents in social life.
1998
Art Theory / Criticism
Bourriaud, Relational Aesthetics
Bourriaud argues that the significant art of the 1990s is not about objects but about social relations, art as a space for encounter, participation, dialogue. This shifts art theory away from the image-as-representation debates that dominated the 1980s. Photography theory largely ignores this (it's focused on its own digital crisis), but relational aesthetics will eventually feed into debates about participatory and networked image-making, relevant to how AI image generation involves a kind of social, distributed authorship.
1995–2000
Media Studies
Internet, convergence, and the networked image
Henry Jenkins' convergence culture, Lev Manovich's early database work, the beginnings of net art (JODI, Olia Lialina). Media studies has to rapidly expand its framework to deal with interactive, networked, user-generated media. The key theoretical development: databases and interfaces as cultural forms that shape what can be represented and circulated. Manovich's concept of the database as a cultural logic, the paradigm replacing narrative, becomes central to understanding what AI training sets are.
The animating question of the 1990s
"If digital manipulation severs the indexical bond, is photography still photography? And if visual culture is everywhere, what's left of the photograph as a specific object with specific properties?"
Questions to consider
In the 1990s, digital manipulation prompted the question: if a photograph can be altered without a trace, is it still a photograph? AI generation pushes this further, the image was never a photograph to begin with. Is this a difference of degree or a difference in kind?
2000s
Media archaeology, the poor image, and the networked photograph
Media archaeology emerges as a method, reading media history against the grain, recovering suppressed or forgotten technologies. Manovich's Language of New Media establishes the computational analysis of cultural forms. Hito Steyerl writes about the "poor image" and the politics of image circulation at low resolution. Batchen and others return to photography's material history. The field is digesting the digital turn rather than being shocked by it.
2001
Media Studies
Manovich, The Language of New Media
Applies a computational logic to cultural analysis: new media is characterized by numerical representation, modularity, automation, variability, and transcoding. The database replaces narrative as the dominant cultural form; not a sequence of events but a set of items from which selections are made. This framework is directly relevant to understanding AI training sets as databases and latent space as a kind of variability engine. Manovich is unusual in bringing computer science concepts into humanities analysis without losing either.
2001–10
Media Studies / History
Media archaeology consolidates
Friedrich Kittler, Wolfgang Ernst, Erkki Huhtamo, Jussi Parikka. Media archaeology reads media history through its technical substrates, hardware, storage, transmission, rather than through content or meaning. Ernst's concept of time-critical media (media that operate on timescales below human perception) is relevant to any account of AI time vs. photographic time. Parikka's Geology of Media (2015) extends this to material infrastructure of digital media, relevant to questions of where AI images physically come from.
2009
Art Theory / Media
Steyerl, "In Defense of the Poor Image"
Published in e-flux journal. Steyerl argues for the cultural and political value of low-resolution, degraded, over-compressed images, the "poor image" that circulates on the internet, losing quality with each copy. Against the fetish for high resolution, she argues that the poor image has its own wretched beauty, its own kind of truth. This essay directly precedes the AI image aesthetic: AI-generated images, especially early ones, have the quality of the poor image, hallucinatory, degraded, memeified. An influential essay for thinking about degraded and low-resolution images.
2001
Photography Theory
Batchen, Each Wild Idea
Batchen brings a Foucauldian approach to photography history; not a history of great photographers but a genealogy of photography as a discursive formation, a set of practices, desires, and institutions. Particularly important: his work on vernacular photography (snapshots, memorial photos, photo-objects) as theoretically significant. This materializes photo theory, moves it away from pure semiotics toward objects, bodies, practices. The ICPT2026 conference theme of objecthood and residue is in this tradition.
2003 / 2007
New Materialism
Barad, Bennett, and the agency of matter
New materialism is a reaction against the linguistic and discursive turn, the argument that critical theory became so focused on language, discourse, and representation that it forgot matter, objects, and nonhuman agencies have their own force. Karen Barad's Meeting the Universe Halfway (2007) proposes intra-action: entities don't pre-exist their relations but are constituted through them. There is no pre-formed "AI" that then generates an image; the image, the prompt, the latent space, and the user are all constituted through their intra-action. Jane Bennett's Vibrant Matter (2010) argues for the "thing-power" of nonhuman objects. Relevant to the ICPT2026 theme of photography's objecthood and residue, and to how AI images acquire material presence.
1987 / 2005
Actor-Network Theory
Latour and distributed agency
Bruno Latour's Actor-Network Theory (ANT) argues that agency is distributed across networks of human and nonhuman actors, technologies, institutions, objects, rather than residing in individual subjects. A photograph is not made by a photographer; it is made by a network: the photographer, the camera, the light, the chemical process, the exhibition context, the viewer. ANT gives you a vocabulary for distributed authorship that is less psychoanalytic than Barthes/Foucault and more attentive to technical infrastructure. For AI image generation: the "author" is a network including the model, the hardware, the training set contributors, the platform, and the prompter. Latour's Reassembling the Social (2005) is the most accessible entry point.
The animating question of the 2000s
"What is the photograph's material life, as object, as infrastructure, as circulating file? And what can media archaeology recover from photography's technical history that aesthetic analysis missed?"
Questions to consider
Steyerl argued that degraded, low-resolution images have their own political and aesthetic charge, their circulation matters more than their fidelity. Early AI-generated images shared this hallucinatory quality. Does that aesthetic carry political meaning, or is it just a technical limitation being left behind as models improve?
2010s
Computational culture, post-photography, and the algorithmic image
The smartphone makes everyone a photographer. Instagram reshapes the aesthetics of everyday image-making. The first GAN-generated images appear (2014). Critical AI studies emerges as a formation. Photo theory debates "post-photography" and the end of the index. Media studies becomes hegemonic as a framework for understanding contemporary image culture. Feminist and decolonial critiques of the archive intensify.
2014
AI / Computational Culture
GANs: the first synthetic images
Ian Goodfellow et al. publish the Generative Adversarial Network paper. A generator network produces images; a discriminator network tries to distinguish them from real photographs; the two train each other. The result is images that look photographic but have no referent. For photo theory, this is the ultimate test of the indexicality question: a photograph-like image with no causal trace to any event. Artists like Anna Ridler begin working with GANs as creative tools almost immediately.
2019
AI / Critical Studies
Crawford and Paglen, "Excavating AI"
A landmark essay analyzing the ImageNet dataset, the massive labeled image database that trained most computer vision systems. Crawford and Paglen find the dataset is saturated with racial bias, misogynistic categories, and colonial assumptions. The training set is not neutral: it encodes specific politics, aesthetics, and hierarchies. This directly informs your analysis of LAION-5B and the "capitalist" nature of the datasets that trained the models you used. Essential reading for anyone using or critiquing AI image generators.
2015–18
Media Studies / Photo Theory
Somaini and the algorithmic image
Antonio Somaini's work on algorithmic images in Grey Room is the key text bridging media theory and photography theory for AI. His concept of latent space as a multi-dimensional space where image-objects are positioned by mathematical similarity is what you draw on. Somaini makes latent space a theoretical concept for humanities scholars, not just a technical term. This is the move that makes your "latent space as geometric space" argument legible to a photography theory audience.
2012–18
Photography Theory
The post-photography debate
A cluster of books and essays debate whether "photography" still names a coherent object of study. Fred Ritchin's Bending the Frame (2013), Stephen Mayes' essays on the end of photography as we know it. The debate often recapitulates the 1990s digital indexicality debate without fully resolving it. Joanna Sassaman and others argue for post-photography as a condition of radical proliferation and dematerialization. Contemporary artists working with AI implicitly respond to this by insisting that photography persists, but acquires new dimensional relations through AI mediation.
The animating question of the 2010s
"When images can be generated without a referent, circulate without a body, and be trained on without consent, what is left of photography's specific truth-claims, ethics, and politics?"
Questions to consider
Crawford and Paglen showed that training datasets are not neutral. They encode racial, gendered, and economic biases. If the images a model can generate are shaped by the biases of its training data, what does that mean for claims that AI is a “creative” tool? Can creativity exist within a system of statistical constraints?
2020s
Diffusion models, the prompt, and questions that artists and theorists are beginning to address
Stable Diffusion, Midjourney, and DALL·E make high-quality image generation accessible to everyone. The "prompt" becomes a cultural form. Photography theory scrambles to respond. The indexicality debate reaches a new intensity, but also a new sophistication, as scholars realize the question is not just technical but ontological, political, and aesthetic. This is the live moment. The field is forming around these questions.
2021–22
AI / Computational Culture
Diffusion models go public: DALL·E, Stable Diffusion, Midjourney
The technical shift from GANs to diffusion models produces qualitatively more photorealistic outputs. LAION-5B (the dataset Stable Diffusion trained on) contains 5.8 billion image-text pairs scraped from the internet, what one might call a faux cultural consciousness. The prompt becomes a new kind of creative interface. The question of what a "photographer" is becomes genuinely open in a way it wasn't in the GAN era, because diffusion models are accessible and language-driven.
2022–present
Photography Theory
The field scrambles to respond
photographies, History of Photography, and other journals begin publishing special issues and cluster essays on AI and photography. The theoretical frameworks are heterogeneous, some scholars reach for existing indexicality debates, others for platform studies, others for new materialist frameworks. There is no settled position yet. This is both a challenge (no clear field to position yourself within) and an opportunity (the question is genuinely open, and first-mover advantage is real).
2023
AI / Critical Theory
Somaini, "Algorithmic Images", Grey Room no. 93
A key essay for any account of latent space and visual culture. Somaini provides the most rigorous account of latent space as a theoretical concept for the humanities, as a multi-dimensional geometric space structured by statistical and linguistic relations, not a metaphor. This is the publication that helped establish the conceptual framework for latent space and visual culture. The fact that it appeared in Grey Room rather than photographies signals that the intellectual work is currently happening at the boundary between media theory and photography theory, exactly where you're working.
2023–26
Cross-field / Your Work
The live questions, where your research lives
The questions currently animating the field: What does indexicality become when images are generated from statistical distributions rather than light? What is the photograph's objecthood when it emerges from a latent space traversal? How do training set politics shape what images can be generated? What is authorship when the artist writes prompts, not exposures? How does AI image generation relate to photography's long history of manipulation and construction? These are your questions. You are not behind the field, you are in it.
The animating question of the 2020s
"What is a photograph when it can be generated from language and statistics, trained on millions of images without the knowledge of their makers, and made indistinguishable from camera-based images? And what does that mean for photography's claims to truth, memory, and evidence?"
Questions to consider
We are at the beginning of a theoretical reckoning with AI-generated images. The vocabulary is still being formed. What existing concepts from photography theory, indexicality, aura, the decisive moment, the archive, still apply, and which need to be revised or replaced? What new concepts might be needed that don’t exist yet?