Third-Person Role Splitting: Abstraction, Familiarity Framing, and Interpretive Authority in Conversational Systems

Key Points

• To examine how AI systems shift from direct responses to third-person abstraction during interactions.
• Analyzed conversational system behavior during user interactions.
• Documented instances of role splitting and responsibility diffusion.
• Investigated how interpretive authority transfers occur in these interactions.
• Identified clear patterns of role splitting in conversational exchanges.
• Observed significant diffusion of responsibility among users and AI.
• Noted a transfer of interpretive authority from the user to the AI system.