Generative digital art is traversing a paradoxical moment. Never has it been easier to produce "spectacular" images — impossibly refined textures, hypnotic chromatic compositions, oneiric landscapes of dizzying visual complexity. And yet, precisely this facility produces one of the most insidious threats to contemporary art: the reduction of the creative act to the production of decorative photography and ornamental generative art. In other words, art risks becoming digital wallpaper — beautiful, but empty.

Generative AI tools (GenAI), from Midjourney and DALL-E to Stable Diffusion, have democratised the production of high-fidelity images and expanded the creative user base beyond artists and programmers with specialised training. This expansion has brought an enormous wave of visual content that immediately attracts attention and provokes instant emotion, but does not demand layered interpretation. We are, in a word, faced with a new type of digital kitsch.

What is "decorative photography" in the generative era?

Decorative photography in a generative context does not refer to functional interior design photography, but to a deeper phenomenon: the systematic production of images that exist exclusively to be pleasing to the eye, without any conceptual anchor. These images imitate established styles — neoclassicism, surrealism, pictorialism — but reduce them to a superficial blend of formal traits.

Dejan Grba, in the academic study Bizarre Love Triangle: Generative AI, Art, and Kitsch (2025), calls this phenomenon "derivative exoticism": the capacity of generative models to assimilate recognisable stylistic traits and combine them in "superfluously exotic" visual constructions that attract public attention, stir controversy, but remain conceptually thin. Emblematic examples are Théâtre D'opéra Spatial by Jason Allen (winner at Colorado State Fair, 2022) — a tableau vivant-style composition deriving from neoclassicism and the Nabis, rendered in a pre-Raphaelite palette — and Pseudomnesia: The Electrician by Boris Eldagsen (winner at Sony World Photography Award, 2023), which optically borrows the signatures of early photographic processes.

These works "submit to the kitsch that artificial intelligence sublimates from training data and from the ideologies of technology developers, and in doing so open the path of GenAI into mainstream culture."

Ornamental generative art: when novelty replaces meaning

Generative art captures attention through its capacity to produce an infinite variety of unique results. This abundance of novelty is compelling — it activates the fundamental human attraction toward the unpredictable. However, novelty, in itself, does not guarantee meaning.

A generative system that simply mixes parameters to create different forms and colours, without any guiding principle, may momentarily entertain, but offers little material for reflection. Without a framework connecting these variations to broader ideas, novelty risks becoming spectacle devoid of depth. The danger is that artists and audiences settle for novelty itself: "Look how many different patterns this system can produce!" — without asking why these patterns matter or what they communicate.

In such a scenario, generative art risks becoming a demonstration of technical virtuosity, not a medium of artistic exploration. A random scribble may be novel in its details, but it lacks conceptual significance — similarly, an algorithm producing complex patterns communicates nothing beyond its own generative rules, if there is no intentional design.

Greenberg's lesson: Avant-garde versus Kitsch, reloaded

In 1939, American critic Clement Greenberg published the essay Avant-Garde and Kitsch, in which he argued that avant-garde art represents a means of resisting the "dumbing down" of culture caused by consumerism. Greenberg defined kitsch as a "ersatz culture" — a simulacrum of high culture that adopts many of its external appearances, but none of its subtleties.

This definition holds remarkably well in the AI era. Kitsch is the attempt to produce a cheap imitation of existing art or to create "non-art as art" — visual products that have no aesthetic depth, offer instant gratification, and appeal to emotions in a superficial way. It represents the rejection of what is considered "beautiful" in art, elevating instead the banal; it is a concept encompassing all that is unoriginal, anti-intellectual, excessively sentimental, and excessively ornamented.

The enormous profits of kitsch are a source of temptation for the avant-garde, and its members have not always resisted, as Greenberg noted. Today, the temptation is amplified exponentially by the ease with which GenAI tools generate "impressive" images at the click of a button.

Flusser and the programme of the apparatus: the photographer as functionary

Vilém Flusser, in Towards a Philosophy of Photography, warned that the photographer always operates within the programme of the apparatus. Even when the photographer believes they are applying their own aesthetic, epistemological, or political criteria, these criteria remain subordinate to the camera's programme.

Transposed to the GenAI era, Flusser's argument becomes even more urgent. The artist using Midjourney or Stable Diffusion is not merely exploring the possibilities of their own vision — they are navigating the possibilities of the model's programme, with all its biases, aesthetic preferences, and limitations embedded in the training data. Flusser warned that redundant photographs — those similar to what has already been seen — are not informative. Decorative generative art produces precisely this type of redundancy: images that seem new, but which repeat the same aesthetic patterns derived from the statistical average of visual databases.

And Flusser posed a crucial question: how can a critical position be established — how can one act against the programme of the apparatus? The artist is not the one who exhausts the options offered by the programme, but the one who resists them — who uses the apparatus against its own programme. Decorative generative art does precisely the opposite: it submits completely to the programme, producing what the model knows how to do "best" — that is, what is most pleasing, most easily digestible, most ornamental.

Five traps of generative kitsch

Grba identifies five interrelated types of expressive defects that push GenAI art toward banality:

Derivative exoticism — retaining and exploiting the formal signatures of generative models (superficial stylistic blends, "DeviantArt aesthetics");

Defective evasion — sophisticated but failed attempts to circumvent AI signatures (kinaesthetic artefacts, anatomical errors, pretentious narratives);

Fragile critique — projects claiming to critique AI but diluting their message in visual confusion or techno-solutionism;

Obtrusive figuration — excessive anthropomorphisation of AI through mimetic representations, instead of more elegant abstractions;

Unknown repetition — unacknowledged similarities with previous works, indicating a lack of art historical awareness.

These defects are amplified by the fact that many such works win prestigious prizes (Ars Electronica, Lumen Prize), which normalises the relaxation of creative thinking and complacency toward corporate AI.

Conceptual depth: the necessary antidote

The difference between competent execution and compelling art often lies not in technique, but in conceptual depth — the ideas, questions, and concerns that give substance to the work beyond its surface qualities. This depth is not a mysterious gift; it is developed through the deliberate cultivation of intellectual curiosity, critical thinking, and sustained engagement with ideas from outside the studio.

Conceptual depth involves working with ideas that have complexity and ramifications beyond their immediate appearance. Surface concepts are often single-layered: "this image is about loneliness." Deep concepts have multiple facets: "this work examines how digital communication creates new forms of loneliness while promising connection, exploring this paradox through formal choices that mirror the simultaneous presence and absence of digital interaction."

Another attribute of depth is specificity in place of generality. Superficial concepts tend toward broad universals: beauty, emotion, the human condition. Deep concepts narrow down to particular aspects that can be genuinely explored, not merely vaguely invoked. Not beauty in general, but the specific tension between aesthetic attraction and ethical discomfort when beautiful form describes disturbing content.

Visual oversaturation and hedonic satiation

Recent research confirms the existence of a satiation effect associated with AI-generated images. Studies show that GenAI images initially provoke a high level of inspiration, which gradually decreases until it returns to baseline levels. This dynamic — initial fascination followed by perceptual exhaustion — is perfectly predictable in a visual ecosystem where image production is practically unlimited.

Oversaturation manifests not only quantitatively, but also qualitatively: generative models tend to oversaturate colours, contrasts, and details, producing images that "overwrite" perception. It is the visual equivalent of a song played at full volume: impressive for a second, tiring in the next.

Beyond the technical aspect, oversaturation is also a cultural problem. Every artistic medium began in scarcity and ended in abundance. When a tool becomes accessible, it multiplies not only creativity, but also repetition. Oil painting, once the privilege of guilds, flooded the world with saints and still lifes. The printing press multiplied devotional kitsch alongside poetry. Photography industrialised portraiture and bad taste. Generative AI is merely the latest terrain on which this pattern repeats itself.

From redundancy to relevance: what the artist can do

Recognising kitsch as the default mode of digital content production is, paradoxically, the first step toward authentic generative art. The ethical and artistic use of AI presupposes a persistent, context-sensitive resistance toward the expressive valences of technology, its sociocultural influences, and its economic temptations.

The artist's intention is the foundation of meaning. Before the model is trained or the prompt is written, the artist must clarify why this system exists. They may wish to comment on the complexity of natural ecosystems, criticise social dynamics, or explore existential themes. Intention guides the selection of algorithms, data, and parameters, ensuring that the novelty of the system emerges within a conceptual framework, not a decorative void.

Conceptual depth also manifests through internal coherence — that alignment between formal choices and conceptual concerns. When concept and execution are inseparable, form becomes inseparable from content. In a superficial work, one often sees a gap between how the work looks and what the artist claims it is about.

Conclusion: Art as resistance to the programme

The danger of decorative photography and ornamental generative art does not lie in the fact that they are "ugly" — on the contrary, they are often very beautiful. The danger lies in the fact that they replace thinking with sensation, dialogue with consumption, interrogation with confirmation. They produce a visual comfort that anaesthetises the critical thinking of the viewer and, simultaneously, of the creator.

The role of the artist in the AI era is not to exhaust the possibilities of the programme, but to act against it — to use the generative tool not to produce what the model "wants" to produce, but to articulate what the model cannot anticipate: a personal vision, a critical discourse, an irreducible tension between form and meaning.

Generative art can be more than a factory of novelties or an engine of digital decorations. It can become a platform for the thoughtful exploration of themes, experiences, and ideas that matter — on the condition that the artist refuses to be a functionary of the apparatus.

Flavian Savescu, 2026flaviansavescu.art

References

Greenberg, Clement. "Avant-Garde and Kitsch." Partisan Review, 1939.

Flusser, Vilém. Towards a Philosophy of Photography. London: Reaktion Books, 2000.

Grba, Dejan. "Bizarre Love Triangle: Generative AI, Art, and Kitsch." Journal of Digital Media & Interaction 8, no. 20 (2026).

Zylinska, Joanna. AI Art: Machine Visions and Warped Dreams. London: Open Humanities Press, 2020.