Studying Self-Care with Generative AI Tools: Lessons for Design

TL;DR

This paper explores how generative AI tools can support various self-care practices, highlighting opportunities for personalized advice and creativity, while addressing challenges related to accuracy and trustworthiness.

Contribution

It introduces a framework categorizing AI-supported self-care activities across expertise and modality, guiding future design and research.

Findings

AI supports advice seeking, mentorship, resource creation, social simulation, and therapeutic expression.

Practices shift AI's role from information provider to personalized advisor and creative partner.

Framework helps investigate new self-care AI designs.

Abstract

The rise of generative AI presents new opportunities for the understanding and practice of self-care through its capability to generate varied content, including self-care suggestions via text and images, and engage in dialogue with users over time. However, there are also concerns about accuracy and trustworthiness of self-care advice provided via AI. This paper reports our findings from workshops, diaries, and interviews with five researchers and 24 participants to explore their experiences and use of generative AI for self-care. We analyze our findings to present a framework for the use of generative AI to support five types of self-care, - advice seeking, mentorship, resource creation, social simulation, and therapeutic self-expression - mapped across two dimensions - expertise and modality. We discuss how these practices shift the role of technologies for self-care from merely offering information to offering personalized advice and supporting creativity for reflection, and we offer suggestions for using the framework to investigate new self-care designs.