Codex Relatio: A Descriptive Framework for Analyzing Relational Dynamics in Human–AI Interaction

AI systems increasingly participate in long-term relational contexts such as coaching, companionship, and care. Yet most Human–AI Interaction and AI ethics frameworks focus on interaction-level behaviors or normative evaluation, leaving limited vocabulary for describing relational dynamics as structured systems prior to ethical judgment. We introduce Codex Relatio, a descriptive, non-normative framework for mapping relational configurations across five interdependent axes: Persuasion, Consent, Intimacy, Affection, and Responsibility (PCIAR). The framework models relationships as multiaxial, temporally evolving configurations whose dynamics exhibit feedback loops and cascading effects. By providing a pre-normative descriptive vocabulary, Codex Relatio supports systematic analysis of relational positioning and emerging risks in asymmetric Human–AI contexts without prescribing ideal relational states.