AI-enabled recursive reprogramming: a social systems theory perspective on algorithmic ambidexterity

Purpose This paper develops a theoretical framework for understanding how artificial intelligence transforms organizational change through recursive reprogramming of decision premises. Drawing on Luhmann’s social systems theory, it conceptualizes AI as evolutionary communication medium rather than tool. The framework explains how AI enables organizations to modify the logic governing their own modification processes, reconceptualizing ambidexterity from managing ontological tensions to programming observational distinctions. Design/methodology/approach This conceptual paper develops a systems-theoretical framework integrating Luhmann’s theory, particularly autopoiesis, second-order observation, and decision premises, with ambidexterity literature. The framework introduces three levels: conditional/purposive rules, self-modifying programs, and recursive reprogramming (meta-programming). It contrasts systems-theoretical and capability-based ontologies, demonstrating how recursive reprogramming dissolves traditional ambidexterity paradoxes by transforming structural tensions into programmable observational distinctions. Findings Recursive reprogramming operates through self-modifying programs, meta-programming enabling second-order observation, and multi-stability through operational diversity. AI functions as evolutionary communication medium collapsing linear time, integrating incompatible logics, and enabling computational observation. Ambidexterity is reconceptualized from ontological categories to contingent observational distinctions. The framework transforms organizational structure from designed configuration to emergent communication pattern, making paradox itself programmable. Originality/value This paper introduces recursive reprogramming extending Luhmann’s theory to AI-mediated contexts. It reconceptualizes ambidexterity from ontological tension to observational distinction, advancing from first-order ambidexterity (balancing activities) to second-order ambidexterity (programming observation itself). It demonstrates how organizational change becomes reprogramming of decision premises through algorithmic mediation, advancing systems-theoretical understanding of digital transformation.