Lived experience in dialogue: co-designing adaptive support in large language models for youth well-being

Youth increasingly turn to large language models (LLMs) for well-being support, yet existing adaptive strategies often fail to reflect the diverse lived experiences and circumstances shaping youth's daily needs. This work contributes human-centered foundations for adaptive user modeling and personalization in youth well-being LLM systems through community-grounded dialogue strategies. We conducted a participatory study with youth, parents, and youth care professionals (total N=38). Thematic analysis highlighted how community stakeholders prioritized dialogic scaffolding that supports youth agency. In particular, stakeholders emphasized adaptive LLM strategies such as clarifying support priorities, meaning-oriented reflective inquiry, exploring personal narratives and future aims, and preserving decision-making. Based on these insights, we developed dialogue extracts directly from stakeholder quotations to inform the design and future alignment of LLM-based well-being systems. By framing adaptive support as an ongoing dialogic process, this work offers human-centered guidance for tailoring support to youths' evolving contexts, priorities, and lived experiences.