The Goddess of chance: randomness from ancient “automaton” to AI-art

This paper examines how randomness contributes to the perception of creativity and autonomy in both natural and artificial systems, particularly in art practices. From ancient accounts of accidental inspiration to modern aleatory practices, the use of chance has long served to unsettle habitual patterns and generate new aesthetic possibilities. Historical automata already demonstrated how mechanical structures could transform irregular events into coherent forms, producing effects that appear spontaneous despite their construction. The paper extends this genealogy to contemporary generative-AI systems. Modern machine-learning models rely on stochastic components that render their behaviour partly unpredictable. Uncertainty in data and learning dynamics contributes to outputs that surprise users while maintaining coherence. This structured unpredictability is central to why such systems are often perceived as creative or quasi-autonomous. The paper then discusses the tension between automation and autonomy: both notions describe processes that "happen of themselves" (as in Aristotle's idea of automaton), yet they appear to be opposed. From this discussion, a typology of different uses of chance and random phenomena is proposed, distinguishing different levels at which stochasticity can be harnessed by (artificial and natural) systems. Ultimately, the paper argues that attributions of creativity and agency depend on a balance between indeterminacy and structure.