Submedial Space

In the AI Story

The visible surface of AI is the polished output: the well-structured text, the functional code, the persuasive argument. The hidden depth is the billions of tokens of human-generated text from which the machine extracted the patterns it reproduces. To understand the output, one must understand the archive — its composition, its biases, its exclusions, its characteristic blind spots. But the smooth surface of the output systematically discourages this excavation. The surface says: I am sufficient. I am the thought, complete and self-contained. There is nothing behind me that you need to see.

The concept has a second application that becomes crucial when the authorship question is raised. The author of an AI-assisted text is, in one sense, a guarantor of material she has not fully examined — which is to say, she is sincere about some of the text and uncertain about the rest, and the boundary between the sincere and the uncertain is invisible to the reader. The submedial space of the text is the author's accumulated experience, her biographical specificity, the angle of vision only she could bring to the material. The machine can generate the surface. Only the human holds the depth. Authorship, reconceived through Groys's framework, is not surface control; it is depth.

This reconception does not resolve the authorship question cleanly. It leaves open the possibility that the depth is thinner than the author believes — that the biographical specificity she claims is itself a performance, that the judgment she exercises is shaped by biases she has not examined, that the experiences she draws on are less determinative than she assumes. These possibilities persist. But the submedial space framework allows the question to be asked in a form that does not collapse into either triumphalism or despair.

The critical practice that follows from this concept is excavation. The reader who engages critically with AI output is not merely evaluating the surface; she is asking what archive produced it, what biases shape its patterns, what voices are overrepresented, what perspectives are systematically excluded. These questions treat the output as evidence of its submedial space rather than as a finished product to be consumed. They are, in Groys's framework, the only form of reading adequate to the AI moment.

Origin

Groys developed the submedial space concept in Under Suspicion: A Phenomenology of Media (2000, English translation 2012). The book treats every medium as a dialectic between the visible surface and the invisible depth, arguing that media criticism must attend to both or risk being deceived by the surface. The framework's extension to AI in the 2020s required no substantive modification — the concept had been formulated in terms general enough to apply to any system of cultural production.

Key Ideas

Surface and depth are structurally related. The visible is constituted by what the invisible makes possible; to evaluate the surface, one must excavate the depth.

The archive is AI's submedial space. Every polished output rests on a training corpus whose composition shapes what can appear on the surface.

Authorship is depth, not surface. The author is the one who holds the submedial space — the experience and accumulated judgment the machine cannot provide.

Critical reading is excavation. The reader adequate to AI output reads it as evidence of its source archive rather than as a self-contained artifact.