Comprehensive Summary
This study, presented by Dallison et al., examined the application of digital mental health interventions (DMHIs) and Generative AI (GenAI) on helping adolescents living in low-middle income countries overcome anxiety and depression. By administering full day workshops in rural Mpumalanga Province, South Africa that allowed participants (n= 23) aged 15-19 to use GenAI develop content in focus groups. Inductive thematic analysis was applied to analyze the transcribed and translated discussions (Xitsonga to English). A high percentage of adolescents (96%) were reported to have no experience working with GenAI in the past. “Enjoyable” and “Engaging” were words used by a majority of participants (87%) to describe the experience. However, participants needed guidance writing the prompts (78%) in order to receive effective output. Some of the themes identified noted how cheering the experience was, the concern for human-based creativity compared to AI, and how outputs often provided a biased, westernized perspective (57%) unless specifically directed otherwise. Dallison et al. acknowledged the cultural bias embedded within AI models given the training done on western datasets.
Outcomes and Implications
A therapeutic aid for individuals suffering from anxiety or depression in low-middle income countries, DMHIs and GenAI could allow for mental health interventions to become more accessible to the wider public. However, sensitivity to biases would need to be prioritized as the models are developed. When taken into consideration, DMHIs could provide the mental health assistance necessary across the globe.