Beyond Forms and Interpretations

“The question of whether artificial intelligence can be said to make art remains inseparable from questions about how meaning is formed within such systems. Rather than treating artistic output as self-evident, this paper attends to the internal representational processes through which meaning may emerge. Drawing on depth psychology, classical Chinese philosophy, and contemporary research in AI interpretability, it explores whether traces of an AI “unconscious” might become perceptible when machine-learning systems are placed under symbolic constraint. This paper describes the genesis and development of an artistic experiment in which multiple sentence-embedding models are restricted to a fixed symbolic vocabulary derived from the 214 traditional Chinese radicals, producing divergent yet interpretable visual forms. Presented as a preliminary exploration, the paper proposes an artistic method for attending to opacity in machine-learning systems while reflecting on how AI reframes human assumptions about creativity, interpretation, and self-knowledge.”

Keohane presented his paper “Between Forms and Interpretations – The Digital Unconscious” at xCoAx 2026, the 14th Conference on Computation, Communication, Aesthetics & X in Turin.

The paper was published in the proceedings for 2026 and can be accessed here.