Most-historical Apparatus : “AI” and the Crisis of Linearity
This paper analyzes machine learning-based services such as LLMs and AI-image generation through Vilém Flusser's account of linearity, causality, and rationality as historically specific cultural techniques. It argues that these technological applications do not suspend linear thinking but operationalize it at scale through ramified, sequential procedures that remain strictly causal. The apparent non-linearity of probabilistic generation of media artifacts such as text, image and video belongs to the level of experience, not to the functioning of the apparatus. Rationality persists within these technological media as program. The contemporary crisis of rationality is therefore not its disappearance but its withdrawal from conscious critique into automated systems controlled by institutional power. The paper argues that the foundational democratic intuitions which emerged with the development of linear writing and linear rationality have been historically constrained by extra-rational exercise of power. Only through the radical democratisation of that power can the emancipatory potential promised of hyper-rational so-called “AI" be realized.