Communicating with the Unfamiliar and the Incommensurable: Generative AI, Neurodivergent Authorship, and the Transformation of Epistemic Form

This paper examines authorship under the conditions of generative AI and challenges the assumption that authorship can be determined through the immediate provenance of individual words, sentences, or other elements of a work. Instead, it proposes understanding authorship as an Eigenzeit relation: as an irreversible relation in which a form emerges from a historically situated process of form formation, enters a shared world as a setzung, and thereby at the same time changes the future possibilities of the one who carries this setzung. From this perspective, text genesis and work genesis are not identical. A machine can generate, transform, translate, reorganize, or materially realize elements of a work without thereby necessarily occupying the same authorial position as the human whose formative history is carried forward through the work. For this, the paper develops the concepts of the Strich, the decision of difference, and operator authorship, as well as the distinction between execution provenance, work provenance, and the authorial act. A central focus lies on neurodivergent authorship. For authors whose cognitive, linguistic, or epistemic organization does not correspond with dominant conventions of communication, generative AI can function not only as a writing tool, but as an externalized foreign Raster through which an already existing form can be transformed, tested, misunderstood, and returned to the author. Building on the concept of Interface Text, it is argued that transformation can, under certain conditions, not diminish human authorship but protect it—namely when communicative accessibility is increased without eliminating that formative difference which made mediation necessary in the first place. Generative AI therefore appears neither as a neutral tool nor automatically as a co-author. It can become an epistemic probe, a site of self-alienation, resistance, and recursive form formation. At the same time, its tendencies toward smoothing, normalization, premature closure, and the internalization of dominant Raster create new risks for the integrity of authorship. The paper further distinguishes between work provenance and cultural provenance in order to examine training data, influence, appropriation, plagiarism, style imitation, and the sedimentation of earlier works in generative systems. AI produces less cultural impurity. Rather, it changes the density, speed, and invisibility of cultural mediation. The model of authorship developed from this is deliberately substrate-neutral. Current generative systems do not possess authorship in the strong sense developed here, because their outputs do not yet become part of an irreversible world-history of their own whose consequences they themselves must carry. The question of machine authorship would only arise anew in a stronger sense once machine setzungen begin to irreversibly change the future possibilities of the machine itself within an Eigenzeit of its own. The paper therefore concludes that authorship should be understood neither solely as purity, nor as manual execution, nor as protection against change. Authorship instead protects the continuity of an Eigenzeit through transformation.