Designing Chatbots and Social Robots for Mental Health to be Culturally Competent and People-Centered

Abstract

AI is increasingly being deployed for patients in mental health settings. While attention has been given to ensuring AI for mental health is person-centered, no study we are aware of has considered how understandings of ‘person-centered’ healthcare vary across cultures or examined the implications for designing AI for mental health. This paper fills this critical gap. We ask, how do understandings of ‘person-centered’ healthcare differ in Japanese and Western societies and how do these differences manifest when AI is used for mental health? To answer these questions, we use Japanese and Western philosophical frameworks to analyze cases involving conversational AI and social robots for mental health and identify salient differences. The analyses suggest the need to ensure AI for mental health is culturally calibrated. We translate this requirement to actionable guidelines that recommend designing AI for mental health to be culturally competent, people-centered, able to keep humans ‘in the loop,’ personalized, privacy respecting, relationally oriented, and decolonized. More research is needed to develop metrics for culturally responsive AI, test guidelines in diverse settings, and consider their application to other healthcare areas.