The Revolution of Artificial Intelligence in Photography—Friend or Foe?
26 • 03 • 16Benedek Szabó
In discussions about artificial image-generating processes, a common question arises: Can images created by AI be considered or approached as photographs? A defining characteristic of these generated images is their attempt to mimic the visual qualities of photography. The compositions seen in artificial images—including their color schemes, visual aesthetics, and overall execution—often resemble the visual language typically associated with photographs. However, viewers must be aware of several important differences. One key distinction is that artificial images do not capture the appearance of the external world; instead, they generate visual scenes through algorithms.[1]
This distinction illustrates that various ways of viewing become possible. Photographic theory has often emphasized that the true power of a photograph—and its uniqueness among other forms of imagery—lies in the fact that the scene depicted actually existed in front of the camera. While the production of a painting or graphic work is not bound to the existence of the spectacle it represents and can depict an image independently of it, the photographic apparatus must necessarily be present. This notion is especially true in the case of traditional, film-based photography, where the film can genuinely “see” the scene it records, establishing a particularly close relationship with the subject of representation. This connection between the finished image and its referent can be most clearly demonstrated in photography: it can preserve memories of the past precisely because it maintains a real relationship to what it depicts. The referent is always distinct, and what appears on the surface of the photograph must necessarily have been present in front of the lens at the time of exposure; without this presence, light-based representation is impossible.[2]
Certain philosophies of art interpret this process of creation as a form of intact recording, often likening it to the making of a death mask. The analogy holds insofar as the creation of a mask likewise begins with a negative imprint, from which the “positive” form is subsequently cast.[3] This logic closely parallels the photographic process, in which a negative is transformed into a positive image. The direct relationship between the photograph and its subject is further articulated through the concepts of the referent and the punctum—the element that pierces or wounds the viewer. In this context, the image detail that guides the gaze does not merely attract attention or elicit an emotional response; it has the capacity to shock or even figuratively injure the spectator.[4] Images, when approached as artistic objects, can therefore be interpreted through specific frameworks.
Beyond this representational aspect—primarily associated with art photography—it is worth turning toward everyday photographs. These images are often created not with artistic intent but for purposes like documentation or preserving memories, and they operate through fundamentally different processes than generated images. Given their immediacy, photographs are particularly well-suited to preserving memories. It is therefore unsurprising that in everyday life, photographs often function as more or less objective custodians of memory, or as triggers for recollection—one need only think of family photo albums.
And what does all this have to do with images created by artificial intelligence? Essentially, nothing. Precisely for this reason, many interpretative frameworks developed for photography cannot be applied to artificial images, as these works are not created with light and do not depict the external world. Consequently, the crucial points of access for an aesthetic evaluation of photographs are not applicable to them. Yet it is precisely here that a possible approach to artificial images begins to emerge. Most photographic theories focus on phenomena that exist in photographs but are absent in images created by AI. This suggests that the two forms of imagery are fundamentally different. The challenge, though, is that this distinction becomes harder to maintain as image-generating algorithms continue to advance. These algorithms are now not only capable of creating independent visual works within the aesthetic realms of films and comics, but they can also alter or manipulate specific details in photographs.
0103
Boris Eldagsen: PSEUDOMNESIA III. Psychoanalysis Gone Wrong
The question of image authenticity dates back to the invention of light-based image recording. Anyone studying the history of photography as an art form will encounter numerous examples of manipulated photographs. Even in the early years of photography in the nineteenth century, there were instances of intentional image manipulation designed to deceive viewers.[5] Harmful effects arise when the artificial enters the domain of the real. A notable example occurred two years ago at one of the most prestigious photography competitions, where the jury selected an AI-generated image as the winner. At the Sony World Photography Awards 2023, an award was presented to an image created by Berlin-based artist Boris Eldagsen using artificial intelligence. The artist declined the award, as his intention was to initiate a dialogue on the issue. Unfortunately, amid ambiguous reactions and the jury’s silence, this aim was only partially realized.[6]
Artificial images and photographs can thus become entangled in specific contexts, potentially leading to the displacement of traditional photographs from their established domains as AI-generated images proliferate.[7] Several art-historical and philosophical approaches conceive of the broad family of images as a system in which paintings coexist with works produced through mechanical reproduction—such as lithographs, zincographs, or photographs. The relationship between photographic reproduction and AI-generated images, however, diverges at the point where the aim shifts from artistic intent to deception. The primary problem with content manipulated or generated by artificial intelligence, such as deepfakes, is that manipulation is always evident. Images or moving images created in this way can easily function as instruments of malicious deception or fraud.
The question of whether artificial creations can be classified as artistic products remains open. This classification depends on whether they have the same art-historical or genre-based embeddedness that traditional artworks possess. Currently, there is limited scholarly literature on this topic; while it has begun to attract attention, the genre itself is still relatively new. This attribute alone, however, is not a sufficient reason to dismiss generated images. Photography itself was once a similarly new medium, whose artistic potential was far from evident in its early years. Over time, photography has gained acceptance as an art form.[8] It is therefore not implausible that works created using artificial intelligence will find their place, eventually establishing their path after an initial period of uncertainty. The increasing evaluation of these images in separate competitions, clearly distinguishing them from photography, already suggests this trend.[9]
The present essay primarily uses art-theoretical and aesthetic approaches to illuminate the issue within its cultural and historical contexts. While these frameworks allow for the establishment of genre-based and technical distinctions, it is equally important to consider the economic implications of artificial image-generation processes—most notably, the impact that the growing availability of freely accessible images may have on professional photographers. If newspapers and advertising agencies increasingly turn to such materials instead of commissioning work from photographers who rely on paid assignments, the negative consequences are evident. Nevertheless, it also becomes clear that photography as an art form is not at risk. In both its creative capacity and everyday practical applications, photography remains irreplaceable.
[1] A particularly illustrative example of this is an artificial intelligence–based “camera” capable of generating an image of a given location based on GPS coordinates and basic settings. Introduced in 2023, the device—named Paragraphica—operates by collecting data on site, which it then incorporates into the process of artificial image generation. The machine determines its precise geographical position, analyzes the corresponding weather conditions and the current time of day, and generates an artificial visual scene accordingly.
Source: Szabó, Benedek. “Mi történik, ha a kép a látvány helyébe lép? – Az AI felkínálja a fényképezőgépek alternatíváját” [What Happens When the Image Replaces the View? – AI Offers an Alternative to Cameras]. Punkt.hu, 1 July 2023, https://punkt.hu/2023/07/01/mi-tortenik-ha-a-kep-a-latvany-helyebe-lep-az-ai-felkinalja-a-fenykepezogepek-alternativajat/
[2] Barthes, Roland. Camera Lucida: Reflections on Photography. Translated by Richard Howard, Hill and Wang, 1981.
[3] Földényi F., László. A guillotine hosszú árnyéka [The Long Shadow of the Guillotine]. Jelenkor, 2023, p. 129.
[4] Barthes, Roland. Camera Lucida: Reflections on Photography. Translated by Richard Howard, Hill and Wang, 1981.
[5] The authenticity of a photograph believed to be the first image of a tornado, taken in 1884, has been questioned. It has been revealed that this image was actually a composite, created by combining details from multiple photographs, making it a manipulated image. In contrast, a recently discovered photograph taken in Kansas in April 1884 meets all professional criteria and is now considered the earliest authentic photograph of a tornado.
Source: Szabó, Benedek. “Nem az a legelső tornádóról készült felvétel, amelyről eddig azt gondolták” [It Is Not the Earliest Photograph of a Tornado, as Previously Thought]. Punkt.hu, 4 Feb. 2025, https://punkt.hu/2025/02/04/nem-az-a-legelso-tornadorol-keszult-felvetel-amelyrol-eddig-azt-gondoltak
[6] Szabó, Benedek. “A mesterséges intelligencia képe nyerte meg a Sony World Photography Awards egyik kategóriájának fődíját” [An Artificial Intelligence Image Won the Top Prize in One Category of the Sony World Photography Awards]. Punkt.hu, 21 Apr. 2023, https://punkt.hu/2023/04/21/a-mesterseges-intelligencia-kepe-nyerte-meg-a-sony-world-photography-awards-egyik-kategoriajanak-fodijat/
[7] Szabó, Benedek. “Nevezhető fényképnek a mesterséges intelligencia által létrehozott kép?” [Can an Image Created by Artificial Intelligence Be Considered a Photograph?]. Punkt.hu, 25 Apr. 2023, https://punkt.hu/2023/04/25/nevezheto-fenykepnek-a-mesterseges-intelligencia-altal-letrehozott-kep/
[8] Sevcsik Hefelle, Dr. Fényképészet [Photography]. Műszaki, 1987, p. 407.
[9] Kéri, Gáspár. “Újabb mérföldkő az AI által generált képek történetében – Fotópályázatot írtak ki a promptológusoknak” [Another Milestone in the History of AI-Generated Images – A Photography Competition Announced for Promptologists]. Punkt, 5 Nov. 2023, https://punkt.hu/2023/11/05/ujabb-merfoldko-az-ai-altal-generalt-kepek-torteneteben-fotopalyazatot-irtak-ki-a-promptologusoknak/
