Exploring Generative AI in Digital Mental Health: Adolescent Perspectives from Rural South Africa on Culturally Relevant Content Creation (Preprint)

BACKGROUND The Kuamsha app was originally developed for a two-arm randomised controlled pilot trial (the DoBAt study), which assessed the feasibility, acceptability, and initial efficacy of a digital choose-your-own-adventure style serious game that delivered behavioural activation (BA) therapy to adolescents with depression in rural South Africa. OBJECTIVE This qualitative study explored the role of generative artificial intelligence (AI) as a novel method of developing engaging, relatable, and relevant digital mental health content for adolescents in rural South Africa. METHODS Through interactive, exploratory workshops and focus group discussions, adolescents compared stories, images, and songs created by generative AI to those in the Kuamsha app, a digital mental health intervention developed without AI. RESULTS Inductive thematic analysis revealed three themes: ‘Use of generative AI tools in a workshop’, ‘Reflections on the creations and comparison’ and ‘Thinking towards the future’. Adolescents generally showed a preference for AI-generated media compared to the Kuamsha app media, and the creative process was an important aspect of the process for adolescents. CONCLUSIONS This study highlighted current biases of existing generative AI tools while also demonstrating the significant potential of generative AI to enhance ‘real-time’ co-design of digital mental health interventions by incorporating more culturally relevant and personalised content.