Send message to Authors

Paper short abstract

There are AI/ML projects that aim to improve efficiency and inclusivity for small farmers but raise questions about governance and equality. This paper explores Zimbabwe agric apps of AI, algorithmic bias and data colonialism. It use interviews with farmers and developers to evaluate training data.

Paper long abstract

AI and ML are becoming integral to digital agriculture projects in Sub-Saharan Africa. Zimbabwe has seen the deployment of AI-based crop management advisory systems, Remote Sensing Systems, and Credit Scoring Systems by donor agencies and the corporate sector, aimed at improving “efficiency” and “inclusion” for small farmers. However, critical questions about power, knowledge, and inequalities remain. It examines the political economy of agricultural AI in the Zimbabwean context through the lens of algorithmic bias and data colonialism. Two Agri Tech projects—a non-governmental organisation-led and a private venture—that employ agricultural AI are used to assess how agrarian data is harvested and constructed by agricultural AI platforms, and their eventual impact on those who benefit and those who are left out of agricultural productivity data systems, especially in Zimbabwe. Typically, data about small farmers, women farmers, communal landowners, and mixed crop and Indigenous farming systems in Zimbabwe remains invisible to such data systems.

Interviews with farmers, developers, and government representatives, discourse analysis of technical and policy texts, and a critique of existing training datasets are used. It contends that AI systems are not objective. Still, reflections and reproductions of agrarian hierarchies highlight emerging criticisms, such as small farmers’ resistance to credit rating applications and the demand for transparency regarding the outputs of artificial intelligence. It argues that emphasis should be placed on local knowledge systems in digital agriculture. It addresses concerns related to the coexisting visions of digital agriculture, exemplifying how AI can exacerbate inequalities unless transformed from the bottom up.