Psychological Well-being and Human-centered AI

Research Overview

In these projects, my research group and I are investigating how AI systems designed to provide emotional and psychological support shape human relationships and the fundamental need for relatedness. Drawing on psychological theories, Human-Computer Interaction (HCI), and Responsible AI, our research examines the risks of artificial intimacy/relatedness and relational displacement of human-connections in human-AI interactions. We propose a relationship-centered approach to AI design, alongside a computational framework to evaluate whether LLMs foster dependency on AI or encourage meaningful real-world human connections.

Research Objectives

1. Investigate how emotionally supportive AI systems influence human relatedness, interpersonal relationships, and psychological well-being.
2. Conceptualize Relational Displacement in AI-mediated therapeutic relationships and develop relationship-centered design guidelines for responsible AI systems.
3. Develop RELATE, a computational evaluation framework to measure the relational orientation of LLMs across multi-turn emotionally supportive conversations.
4. Empirically evaluate how different LLMs and user personas influence inward-facing relational engagement versus outward-scaffolding toward real-world human connections.
5. Inform the design and evaluation of emotionally supportive AI systems that strengthen human agency, psychological well-being, and authentic interpersonal relationships.

Contributions