
6/13/2025
Family Planning AI Chatbot Benchmarking
The START team derived a feasible evaluation method of a large language model (generative AI-powered) chatbot designed to improve access to family planning information among young adults in Kenya and Senegal. This project involved benchmarking the chatbot against global and culturally sensitive best practices for digital health tools in addition to establishing rapid-cycle methodological processes for ongoing refinement and implementation. Through key informant interviews and landscape analysis, the team provided recommendations on how to effectively ensure the chatbot’s accuracy, acceptability, safety, and potential for scale-up across diverse implementation contexts.