This concept concerns the empirical study of affective (emotional-support-oriented) use of AI chatbots, defined as conversations involving interpersonal advice, psychotherapy/counseling, coaching, or romantic/sexual role play, as distinct from an AI system's designed purpose as a general-purpose work tool. The underlying principle is that understanding real-world usage patterns—via privacy-preserving conversation analysis—should ground safety policy and mitigation design, rather than treating anecdotal or headline-driven assumptions as sufficient. This sits within the domain of AI safety and societal-impact research, specifically the sub-areas of user-behavior analysis, safeguards/policy design, and post-deployment monitoring, relating to the broader discipline of responsible AI development and human-AI relationship dynamics.
This concept concerns the empirical study of affective (emotional-support-oriented) use of AI chatbots, defined as conversations involving interpersonal advice, psychotherapy/counseling, coaching, or…