GAIA (Generative AI for Agriculture) and CABI’s governance work

CABI, Digital Green, IFPRIBuilt by: publisher or platformStatus: LiveYear: 2025

Overview

GAIA is a project that aims to make AI-generated agricultural advice more effective, reliable and relevant for small-scale producers in the global South. It is led by the International Food Policy Research Institute (IFPRI), with CABI, SCiO, the University of Florida, Digital Green and the CGIAR Accelerator on Digital Transformation as partners and the Gates Foundation and the UK Foreign, Commonwealth & Development Office as funders.

Phase I, a pilot, tested Digital Green’s Farmer.Chat, a RAG chatbot, in Kenya and India, and examined how CGIAR open-access research and CABI’s proprietary materials could improve its accuracy. Phase II, which IFPRI describes as running to 2027, has CABI leading the work on licensing, governance and ethical guidance.

In a blog post of May 2025, CABI set out six lessons on governance: that flexible licensing builds trust with content providers, that Creative Commons licences may fall short for generative AI, that processes matter as much as policies, that ownership of background and foreground IP and liability need defining early, that attribution should be shown in the chat interface, and that sustainable content sharing needs funding.

Benefits and outcomes

  • CABI reports that its licensing framework, which protected providers’ background IP, set permissions for generative AI use, required attribution in the system and interface, and restricted commercial use without further negotiation, enabled the sharing of 463 expert-validated content items.

Notes and quotes

  • “GenAI holds huge promise for agriculture advisory, but without clear governance, that promise is fragile.” Katherine Cameron, Head of Digital Advisory Tools, CABI.
  • “A policy without a process cannot hold up when requests move fast.” CABI blog.
  • CABI’s blog says restrictive licences such as CC-BY-ND can block the modifications AI systems need and that most Creative Commons licences do not address AI training, so supplementary licences may be needed.

Sources

This entry was written with the help of AI, from the sources listed. Please check the sources before relying on it, and let us know if anything needs correcting.