This white paper presents a socio-technical framework for responsible AI design in African agriculture. Drawing on insights from the AfriCHI 2025 workshop, it argues that agricultural AI must move beyond technical accuracy to account for farmer agency, indigenous knowledge, linguistic diversity, mediated advisory practices, and climate uncertainty.
The paper outlines practical design principles and sector-level priorities for building AI systems that are locally grounded, socially feasible, multimodal, and ecologically adaptive.