arXiv:2601.14401v1 Announce Type: new Abstract: This article introduces Recursivism as a conceptual framework for analyzing contemporary artistic practices in the age of artificial intelligence. While recursion is precisely defined in mathematics and computer science, it has not previously been formalized as an aesthetic paradigm. Recursivism designates practices in which not only outputs vary over time, but in which the generative process itself becomes capable of reflexive modification through its own effects. The paper develops a five-level analytical scale distinguishing simple iteration, cumulative iteration, parametric recursion, reflexive recursion, and meta-recursion. This scale clarifies the threshold at which a system shifts from variation within a fixed rule to genuine self-modification of the rule itself. From this perspective, art history is reinterpreted as a recursive dynamic alternating between internal recursion within movements and meta-recursive transformations of their generative principles. Artificial intelligence renders this logic technically explicit through learning loops, parameter updates, and code-level self-modification. To distinguish Recursivism from related notions such as generative art, cybernetics, process art, and evolutionary art, the article proposes three operational criteria: state memory, rule evolvability, and reflexive visibility. These concepts are examined through case studies including Refik Anadol, Sougwen Chung, Karl Sims, and the Darwin-Godel Machine. The article concludes by examining the aesthetic, curatorial, and ethical implications of self-modifying artistic systems.