Personal informatics (PI) systems, a broad range of technologies that help people provide and obtain knowledge about themselves, have proliferated with the advance of mobile and smartwatch technologies as well as gained interests in mental health and mindfulness. Despite their prevalence, the way existing PI systems communicate with people has not substantially changed for decades; people still capture and consume data in structured ways, often omitting the richer context around the information they provide. 

In this talk, I introduce my recent research at NAVER AI Lab on enhancing PI systems with natural language interactions, incorporating large language models (LLMs). Recent advances in LLMs opened up new opportunities to support natural user interactions, powered by their emergent capabilities of understanding and expressing a variety of human knowledge. I share a series of conversational PI systems that elicit knowledge of interest from users through open-ended and semi-structured conversations while re-structuring the gained information under the hood. I will cover an LLM prompting method for bootstrapping health data collection chatbots, a conversational journaling AI for psychiatric patients, and a chatbot that prompts children to share their emotions and personal events. Going further, I discuss the implications of AIs in personal informatics research and the open challenges of leveraging LLMs for personal informatics.

Young-Ho Kim is a research scientist at NAVER AI Lab, leading the Human-Computer Interaction research group. Before joining NAVER, he worked as a postdoctoral associate in Information Studies at the University of Maryland, College Park (2019-2021). Young-Ho is a Human-Computer Interaction researcher working at the intersection of Personal Health Informatics and Artificial Intelligence. Combining his multidisciplinary knowledge in Computer Science and Visual Communication Design, he has designed and developed computing systems for self-tracking that facilitate people to collect and consume their activity and health data in a flexible manner. He has been recently investigating how large language models can further streamline flexible self-tracking. He has disseminated his research at prestigious HCI and Computer Science venues such as CHI, CSCW, UbiComp, VIS, and DIS. 

Young-Ho received a Ph.D. degree in Computer Science and Engineering (2012-2019) and a Bachelor of Fine Arts degree in Visual Communication Design (2007-2011) from Seoul National University. He is a recipient of the Korea International Postdoc Fellowship supported by the National Research Foundation of Korea in 2019, a Best Paper award at ACM CHI 2023, and an Honorable Mention award at ACM CHI 2021. For more details, see his website at younghokim.net.