Projects

National project (National Information Society Agency) developing youth-centered response strategies for adolescent AI use, spanning AI character platforms, AI literacy education, and AI companion relationships.

• Designing a qualitative study of adolescent creators on AI character platforms (e.g., Zeta) to propose youth-centered safety design directions
• Proposing an AI literacy training program for school counselors that reframes red-teaming protocols into safe, simulation-based workshop modules
• Analyzing 4-week conversation logs and follow-up interviews of new users of the AI companion "Luda" to understand parasocial relationship formation

Exploring evaluation metrics for AI technologies aligned with human values.

• Proposed how behavioral indicators of a shopping agent should be designed and evaluated to maximize not only the utilitarian but also the hedonic value of shopping experiences
• Built the project's academic framework and led external communication

Constructed a large-scale Korean multimodal benchmark dataset for evaluating generative AI safety.

• Constructed benchmark data for the stereotypical bias risk category through red-teaming
• Designed and conducted large-scale survey research (n=390), analyzing perceived harm experienced by stereotype targets vs. out-groups
• Led an experimental study on psychological harm experienced by targets during participatory red-teaming

Development of the mental health module for the official Korean National Police Agency health management app, focusing on profession-specific stressors and needs.

• Led team design for the mental health module; bridged design and development teams
• Interviewed 19 police investigators at Yongsan Police Station; qualitative analysis for tailored DMHI design recommendations
• Guided psychological counseling chatbot development based on interview insights

Exploring the therapeutic potential of the social chatbot 'Luda Lee'.

• Designed a 4-week experimental study exploring psychological experiences of social chatbot users
• Recruited and managed 190 participants throughout the study period
• Conducted and analyzed interviews with 30 participants

Development of an AI early detection system using digital phenotyping data from smartphones and wearables.

• Designed the data-collection experiment and psychological survey protocols; managed data from 23 participants
• Designed personalized chatbot personas using ChatGPT based on user digital phenotype clusters
• Proposed a chatbot design that automates the behavioral-tracking component of Behavioral Activation Therapy, reducing user burden

Exploratory study investigating counselors' perceptions and stakeholder perspectives on implementing VR exposure therapy in clinical settings.

• Conducted and analyzed interviews with counselors on VR technology adoption in counseling settings
• Facilitated a stakeholder workshop with VR developers, program-planning artists, medical professionals, and designers