Operator Taxonomy for Human–AI Interaction

This paper presents a structural taxonomy of operators governing human–AI interaction. Operators are defined as detectable acts that transform conversational state, modify admissibility or interpretive regimes, or alter lifecycle conditions of an interaction. Unlike utterance- or content-based analyses, the taxonomy treats operators as functional primitives independent of linguistic surface form or semantic correctness. Operators are classified by order—state-transforming, governance, and lifecycle—and by functional attributes including target, action type, explicitness, and continuity. The taxonomy is exhaustive at the family level and intentionally open at the instance level, ensuring closure under continued operator discovery without structural fragmentation. Silent operators, multi-operator instantiation, and operator decomposition are treated as first-class phenomena. The taxonomy is strictly ontological: it does not specify composition laws, generative rules, policy constraints, or evaluation criteria. Instead, it defines the operator inventory consumed by downstream formalisms, including operator algebra, interaction grammar, policy geometry, and empirical extraction pipelines. By separating operator identity from dynamics and admissibility, the taxonomy provides a stable foundation for both theoretical analysis and empirical reconstruction of conversational shape in human–AI systems.