Advancing Gender Equity in Digital Agro-advisory through Inclusive Artificial Intelligence (AI): Bias Analysis and Strategic Recommendations from the iShamba Platform Based on Five Years of Female Farmers’ Queries

Nganga K, Grossi A, Wanjau A. 2025. Advancing Gender Equity in Digital Agro-advisory through Inclusive Artificial Intelligence (AI): Bias Analysis and Strategic Recommendations from the iShamba Platform Based on Five Years of Female Farmers’ Queries. AICCRA Working Paper. Accelerating Impacts of CGIAR Climate Research for Africa (AICCRA)

This working paper presents a comprehensive bias analysis of iShamba’s digital agro-advisory service and provides strategic recommendations to gender-inclusive artificial intelligence (AI) that will be integrated to the platform. Using five and half years of farmer query data (year 2020–2025), we identify multiple biases, including gender participation gaps, referral biases in query resolution, language and literacy-related issues, prevalence of very short queries, content duplication, regional usage disparities, and inconsistencies in units/currency references. The Methodology outlines our data-driven approach and discusses data filtering challenges. A Background and Literature Context situates iShamba within broader efforts on gender-responsive AI in agriculture, drawing on AICCRA, CGIAR, and global studies. In Findings, we quantify each bias and visualize, illustrating how women farmers are underserved or mis-served. We discuss how these biases can undermine equitable service delivery, for instance, how language barriers or regional gaps disproportionately affect women and marginalised groups. The Recommendations section links directly to each identified bias, suggesting interventions such as inclusive AI training data, human oversight mechanisms, data standardisation (for units and terms), multilingual support, and proactive inclusion strategies to engage women farmers. By addressing biases and centering gender inclusion, we can enhance the quality and equity of advisory services. This paper serves as a resource for practitioners and researchers seeking to ensure that AI-driven agricultural advisory tools are fair, context-aware, and accessible to all farmers regardless of gender or region.