How farmers learn in the digital age – International Day special

Beyond informal channels, we can observe more formal and structured digital learning platforms dedicated specifically to agriculture. These platforms offer modular coursework, structured curricula, assessments and certifications, sometimes with blended online and in-person training, formalizing digital learning into recognized skills development pathways.

Structured digital learning platforms offer several distinctive advantages over their informal counterparts. They enable standardization: learning outcomes can be clearly defined, aligned with national agricultural priorities, and linked to broader policy objectives such as climate adaptation or food safety compliance. They also offer considerable scalability, making it possible to train large numbers of learners without a proportional expansion of advisory personnel or physical training infrastructure.

A defining feature of structured digital learning platforms in agriculture is the provision of certification, which frequently serves as a key incentive for participation. Proof-of-completion certificates can either document acquired knowledge and competencies per se, or serve as qualifying credentials for accessing finance, subsidies, value chain participation, or regulatory compliance.

Platforms that provide proof-of-completion certificates primarily position certification as evidence of acquired knowledge and competencies. Notable examples documented in the AgriTech Observatory include:

A more instrumental form of certification connects course completion directly to financial schemes. Two examples from Türkiye are illustrative:

Despite the demonstrated potential of structured digital learning, significant challenges remain. Course completion rates frequently remain low, particularly in contexts characterised by wide variation in digital literacy. Connectivity constraints and limited digital skills continue to restrict participation in rural and remote areas. Sustainability is a recurring concern: many platforms remain heavily dependent on project-based or donor funding, creating uncertainty around long-term viability.
 
Most importantly, despite their scale, the majority of structured programmes continue to deliver standardised content to highly heterogeneous learner populations whose farm sizes, production systems, and agroecological contexts differ substantially. Bridging this tension between standardisation and contextual relevance represents a central challenge for the next phase of agricultural digital learning — one to which artificial intelligence may offer partial, though by no means straightforward, solutions. 

©FAO

Toward AI-driven digital learning

Advances in artificial intelligence, particularly in generative AI, are beginning to reshape how people learn across sectors, and agriculture is no exception. Instead of navigating static courses or searching across multiple websites, farmers can interact with conversational systems that respond instantly to specific, practical questions.

AI can make learning “just-in-time”.  When a farmer encounters a problem — an unidentified pest, an unfamiliar disease symptom, or uncertainty about fertilizer application — a well-trained AI system could provide immediate, context-sensitive guidance. Such interactions could take place in multiple languages and formats, reducing the time between problem and response. In this sense, learning would begin to merge with extension services rather than remaining separate from them, with AI acting as a kind of digital agronomist.

Recent developments in generative AI point to an even more transformative frontier for digital learning. New models are capable of converting textbooks, PDF documents and other source materials into multiple personalized formats — such as narrated audio lessons, simplified summaries, quizzes or interactive explanations adapted to the learner’s interests and level. Instead of consuming fixed modules, farmers could engage with dynamically generated content shaped around their immediate needs: short micro-lessons linked to current field conditions, voice-based interfaces that enable participation by low-literacy users, and adaptive explanations that respond to follow-up questions in real time.

Beyond personalization of format, AI holds potential for contextual personalization of content. Knowledge and advisory outputs could vary depending on farm size, crop type, growth stage and local weather or pest risks. Systems might incorporate past learning behavior and farm-level data to refine advice over time. Gender-sensitive design could allow women farmers, who often face time and mobility constraints, to access learning in formats and at moments that align with their responsibilities. In principle, this could make agricultural learning more relevant, inclusive, and effective than ever before.

However, the realization of this vision depends on structural preconditions that remain uneven across Europe and Central Asia. Rural connectivity gaps continue to affect parts of Europe and Central Asia, limiting farmers’ ability to access AI-enabled services. Language coverage remains uneven, particularly for minority and local languages. Training data are not sufficiently localized, and critical data gaps remain.  High-quality, hyper-local, interoperable datasets are essential for accurate and reliable recommendations, yet such data are frequently incomplete, fragmented or inaccessible. Without them, the risk of generalized, inappropriate or even misleading advice increases.

Governance and equity considerations are therefore central. Key policy questions include whether AI-driven learning will reduce or exacerbate digital divides; whether algorithmic systems can meaningfully adapt to highly heterogeneous agroecological conditions; and whether design processes adequately account for marginalized groups, including smallholders and women farmers. Transparency, data stewardship, and human oversight will be critical determinants of trust and adoption.

 ©FAO/David Khelashvili