Leveraging Human-AI Partnership Approach to Unravel Human Psychological Constructs
abstract: Human psychological constructs form the cornerstone of our understanding of cognition, emotion, and behavior, which are crucial to health informatics and HCI studies. However, current psychological construct analyses may lack theoretical grounding, cultural sensitivity, context, and socio-digital tailored interventions. My thesis explores how these human-centered computing methods, in particular human-AI partnerships, can unravel psychological constructs, using mental health stigma as a case study. Specifically, the study takes a multi-faceted approach: designing conversational agents for multilingual, cross-national data collection from over 1,000 participants; conducting human-AI collaborative qualitative analysis to interpret the embodied psychological constructs; modeling causal relationships from human-chatbot conversations to decompose psychological factors and their interplay; dissecting cross-sociocultural variation; and exploring digital techniques for restructuring psychological constructs. The overall goal is to enhance human-centered AI approaches for psychological-construct analysis, foster cultural inclusivity, inform the design of culturally appropriate healthcare technologies, and promote social justice by operationalizing psychological constructs.