Trustworthy Conversational AI Agents for Human Well-being

Conversational AI agents are transforming how people learn, create, and connect. As these systems become more integrated into everyday life, trust depends not only on model capability but also on accuracy, transparency, human oversight, and rigorous evaluation frameworks—all with the ultimate goal of supporting human well-being. The talk consist of two parts: First, I will showcase a series of conversational AI agents developed through both published and ongoing research projects focused on enhancing human well-being across diverse application domains. These case studies will highlight practical lessons for designing and evaluating trustworthy AI systems that balance technical performance with human-centered outcomes. Second, I will introduce our research group's current interests and future directions, as well as the vision and initiatives of the AI programme within the education committee (OLC). Together, these efforts highlight opportunities to advance responsible, impactful, and human-centered AI for both learning and research.