When Collaboration Becomes State: Formation Attribution in Persistent Human-AI Systems

Summary

This paper examines how persistent AI systems come to carry operational configurations shaped by repeated human interaction. It proposes relationship-derived operational state (RDOS) as a category for persistent configurations that materially depend on an interaction history with a specific participant and were shaped by that participant’s evaluative or directive contribution. It outlines three testable conditions for deciding whether such state should be attributed to that participant rather than to passive inference or generic system behavior: counterfactual dependence, material interactional contribution, and downstream operational consequence.

Evidence status: Conceptual proposal with stated testable conditions. This site makes no claim that the conditions have been empirically validated.

Role in the research program: formation and attribution of persistent collaborative state. It sits alongside the Operant Dyad and Persistent Primitive papers, which address how the relationship forms and how its state persists.