Answers are grounded in this paper and cite their sources.
Beyond Prompt Engineering: Role Reconfiguration in Long-Lived AI Conversations is a field report on long-lived, persistently specialized AI conversations within a human-directed research collaboration, with one designed role-reconfiguration intervention embedded in the study. The paper examines role inertia, context imprinting, represented social position, represented social source, externalized organizational memory, and resistance to role transitions that require a model to affirm unverifiable propositions about its own identity, history, or relationships. It distinguishes protocol-supported observations, ancillary preserved observations, retrospective practitioner observations, and hypotheses, and proposes a countable rule for testing role prompts. This Zenodo release contains the multilingual publication edition in English, Russian, and Dutch. AI systems contributed to manuscript development, scientific editing, methodological review, and ontology review. They are identified in the article by function rather than listed as authors; final publication authority and responsibility remain with the human author. Companion documentary volume: A. Tuin, Chronicles of the Prompter, Zenodo DOI: 10.5281/zenodo.23399552.
No takes yet. Share an insight, caveat, or question.
Valerii Krasnianskyi (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: