Dissociation Between Surface Politeness and Social Architecture in Human-AI Dialogue
September 21, 2026
Analysis of over 26,000 dialogues reveals that ChatGPT simulates the surface features of cooperative communication without the underlying social architecture. Findings show that AI-generated morality is preconfigured rather than negotiated, and linguistic convergence declines over multi-turn interactions compared to human-human dialogue.
HOW THIS AFFECTS YOU
●
researcherThis suggests current alignment metrics may fail to capture the absence of true cooperative mechanisms.
●
policyYou should consider that social alignment in models may be a superficial simulation rather than a robust capability.