In its current widespread use, generative AI primarily operates at the affective level: it is optimised for response speed, visual appeal, and compliance with average “like” patterns. The question of whether the technological environment is capable of creating conditions for experiencing the second level—leaving a trace, initiating an encounter with the unrepresentable—remains open.
The emergence of generative artificial intelligence was an event which resonated far beyond the world of art. Technologies capable of creating images, text, music, and video have attracted the attention of political players—states, platform developers, investors—those who have not traditionally been involved in aesthetics, the philosophy of art, or art itself. Behind this attention from regulatory actors lay an idea that philosophy has formulated more precisely than they themselves could: whoever controls the production of images and meaning on a mass scale controls more than just the content market. Controlling the desirable, the beautiful, the ugly, the funny influences the sensory perception of communities, cultures, and civilisations.
To understand the nature of this political-technical-aesthetic connection, let us recall the Greek concept of techne (τέχνη). In ancient usage, techne simultaneously encompassed art, politics, skill, knowledge, and craftsmanship—everything through which a person was believed to enter into meaningful, harmonious interaction with the world, complementing and ordering nature. In the modern era, techne has been reduced to technology—the instrumental rational transformation of reality for the sake of utility. Technology became a means, and the engineer and the artist diverged into incompatible conceptual systems. Today, through the fusion of AI technologies and human design, techne has a chance to return—as an understanding of creative harmony achieved through the materialisation of new images and spaces.
The dialogue between the creator and the generative model reproduces that ancient logic of co-creation with the world that the Greeks conceptualised when they spoke of techne.
However, this return risks not taking place—not only because of a lack of knowledge and skills among ordinary users, but also because of the claims made to generative AI by those who hold power and seek to use force in the struggle for energy production, the extraction of rare earth metals, water, the creation of computer system components, and the AI models themselves.
Aesthetic and Political: From Aisthesis (Αἴσθησις) to Sovereign Perception
The struggle for the right to define “friend/enemy,” “norm/abnormal,” and the mechanisms of power have been examined by many philosophers—Carl Schmitt, Michel Foucault, Roland Barthes. But from the perspective of impressions and sensory perception—which we are interested in through the prism of aesthetics—Jacques Rancière addressed the problem of politics and power in the aesthetic most directly. Having returned aesthetics to its pre-Socratic dimension—close to the very word aisthesis, which denoted sensory perception as a bodily encounter with the world, long before modern philosophy turned aesthetics into an autonomous theory of beauty—he introduced the concept of le partage du sensible, the division of the sensible. Perception, as the philosopher showed, is never carried out in a neutral space; it is structured by a predetermined order, in which it is distributed by who is seen and heard in the general field, whose presence is considered significant, and what is perceived only as noise. Politics, in Rancière’s understanding, is the struggle for the regime of visibility—for the right to be perceived not as noise, but as a subject. He also noted that the aesthetic becomes political not by its nature, but under certain conditions.
To understand why Rancière’s vision of politics and the power of sensory impression takes on particular urgency in the age of generative AI, it is necessary to distinguish between two levels of impression. The philosophical foundations for their distinction were laid by Jean-François Lyotard and Jacques Derrida. Lyotard, developing Kant’s analytics of the sublime, insisted that it is an experience that fundamentally exceeds the capacity for representation and comprehension. In the gap between what happened and the inability to comprehend it, a thought is born. The sublime is not produced to order—it arises where form capitulates to what cannot be formed. Derrida, through the concept of différance, showed that meaning is always deferred, arriving through a trace. A deep impression continues to “work” after the encounter with the object, does not let go, requires interpretation—and in this capacity it acts as the antipode of virality: content optimised for instant affect leaves no trace and does not demand interpretation.
Thus, these two levels:
1. Affective. Engages chemoreceptors and other sensory channels—such as vision, hearing, and tactile sensations—and serves to produce a rapid physiological response, broadcasting mass images designed to capture attention. The impression at this level is predictable, projectable, replicable, and marketable. It is intended not only for oneself, but also for the Other—as a gesture of self-presentation, a tool for satisfying ambitions, a status signal.
2. Experience. An ongoing existential event that is not exhausted at the moment of contact with an object. It is with experience that true reflection begins, including aesthetic reflection. Experience is not produced; it happens.
Between these levels lies the entire space of contemporary creative production—the experience industry. While affective impressions are produced and sold, experience fundamentally defies direct monetisation. According to Rancière, whoever controls the infrastructure for producing first-level experiences at scale controls sensory perception—shaping what is perceived as worthy of attention. This power over the division of the sensible is no less real than legal power.
Generative AI is a “machine,” a “factory” for producing impressions. The developer, by selecting training data and target functions, exercises sovereignty, determining which sensory mode will be broadcast to millions of users. In current mass deployments, generative AI primarily operates at the affective level—optimising for response speed, visual appeal, and conformity to average “like” patterns. The question of whether the technological environment is capable of creating the conditions for a second-level experience—one that leaves a trace, opening an encounter with the unimaginable—remains open. The answer is determined not only by architectural decisions, but also by the sensory mode initially embedded in the system’s foundation.
Thus, the aesthetic, the technical, and the political are intertwined in the struggle for the right to sovereign expression as the right to set the focus of visibility, to define more than just friend/enemy in Schmitt’s sense, but also, in his logic, beauty/ugliness—and to construct trajectories for the creation of meanings and values. This is precisely why states and regulators, developers and investors quite consciously claim a monopoly over creative expression. From this perspective, the struggle for sovereignty in the field of generative AI takes on a dimension that extends beyond technology or economics—a political-aesthetic one.
