`Socheton': A Culturally Appropriate AI Tool to Support Reproductive Well-being

TL;DR

This paper presents 'Socheton,' an AI tool designed to improve reproductive well-being education in Bangladesh by addressing misinformation, cultural language, and community participation, informed by ethnographic research.

Contribution

It introduces a culturally tailored AI-mediated platform for reproductive health education, integrating community and professional moderation based on ethnographic insights.

Findings

Combating misinformation alone is insufficient for community engagement.

Culturally appropriate language is crucial for effective reproductive health communication.

Community participation influences the success of reproductive well-being initiatives.

Abstract

Reproductive well-being education in the Global South is often challenged as many communities perceive many of its contents as misinformation, misconceptions, and language-inappropriate. Our ten-month-long ethnographic study (n=41) investigated the impact of sociocultural landscape, cultural beliefs, and healthcare infrastructure on Bangladeshi people's access to quality reproductive healthcare and set four design goals: combating misinformation, including culturally appropriate language, professionals' accountable moderation, and promoting users' democratic participation. Building on the model of `\textit{Distributive Justice,}' we designed and evaluated \textit{`Socheton,'} a culturally appropriate AI-mediated tool for reproductive well-being that includes healthcare professionals, AI-language teachers, and community members to moderate and run the activity-based platform. Our user study (n=28) revealed that only combating misinformation and language inappropriateness may still leave the community with a conservative mob culture and patronize reproductive care-seeking. This guides well-being HCI design toward being culturally appropriate in the context of reproductive justice with sensitive marginalized communities.