The World Bank proposes creating national databases for agricultural AI

According to a World Bank report, investments in agricultural statistics are proposed to be transformed into digital infrastructure. This will make it possible to support artificial intelligence-based services, such as crop mapping, yield forecasting, pest monitoring, and targeted agricultural advisory services. Nigeria is among 10 African countries where data on farms, crops, production, resources, livestock, and farming practices has already been collected with the support of the bank and the 50x2030 initiative.

The bank notes that satellite imagery, meteorological data, and soil maps alone are not enough for AI to function, as they do not reflect the situation on specific farms. Agricultural surveys can provide the necessary ground-level data. When combined with other sources, such georeferenced data are suitable for training and verifying AI models. A study in Uganda is cited as an example, where the combination of satellite imagery and household surveys improved yield estimates and reduced errors.

For Nigeria, the integration of surveys with satellites, meteorological data, soil maps, and market information will help identify crops, forecast yield and loss risks, and detect threats from drought or pests. Such platforms will provide ministries and researchers with access to systematized data, provided that measures to protect farmers' confidentiality are observed.

At the same time, the World Bank emphasizes that AI readiness requires standardization, documentation, georeferencing, interoperability, and data security. Four key requirements for AI are highlighted: connectivity, computing, context and competence, as well as management and cybersecurity. Agricultural surveys contribute to providing the "context."

The bank recommends starting with practical tasks: mapping, crop forecasting, loss and drought monitoring, pest surveillance, and advisory services. In the future, it will be possible to develop regional cooperation and create a federated African agricultural data architecture, which will allow countries to share standards and methodologies while retaining control over their own information.