In partnership with the Family Planning and Gender Equality Division at the Gates Foundation, the START Team assessed the design and implementation considerations for a generative AI-powered chatbot intended to expand access to contraceptive information among young adults in Kenya and Senegal. The chatbot aims to address gaps in trusted, youth-friendly family planning (FP) resources by leveraging conversational AI to deliver on-demand, contextually relevant guidance in local languages, via multimedia youth engagement platforms “Shujaaz” (Kenya) and “C’est la vie” (Senegal).
The team benchmarked the chatbot against key performance metrics, including data privacy and security, safety, and accuracy. Key metrics domains included response accuracy, hallucination rates, and alignment with national Sexual and Reproductive Health standards (Kenya and Senegal). Additionally, the team employed a rapid-cycle process and documented a tailored implementation model outlining how the chatbot could be iteratively improved through agile data collection and feedback loops. A three-phase framework was proposed to support safe and scalable deployment, including pre-launch validation, real-time monitoring, and post-deployment adaptation. Findings and tools were shared through a final slide presentation, an Excel spreadsheet with benchmarking information, and the rapid-cycle process document. Final recommendations addressed how the Gates Foundation could support ethical, evidence-based expansion of AI-powered tools in family planning programming while maintaining alignment with country strategies, youth engagement principles, and digital safety protocols.
