The Persistent Primitive: Configuring Human-Governed Continuity for Persistent Human–AI Interaction

Abstract

Persistent human–AI interaction can generate accumulated relationship state: retained context, corrections, decisions, constraints, preferences, interpretations, and working history that alter later interaction.

This paper argues that useful continuity requires more than retrieval. It derives requirements for provenance, temporal validity, correction, authority, purpose limitation, revocation, portability, and protection against machine inference hardening into durable identity. For the living, capacitated case, it treats the natural person as the primary design locus because models, providers, applications, representations, and transport mechanisms remain substitutable.