Artificial Intelligence in Agricultural Extension and Advisory Services: A Scoping Review of Applications, Explainability, Trust and Farmer-Centred Adoption
This scoping review synthesizes evidence on AI and explainable AI in agricultural extension, highlighting diverse applications and decision-support tools while emphasizing that successful adoption depends on addressing farmer-centred factors like trust, explainability, and contextual relevance rather than technical performance alone.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
For generations, the bridge between scientific discovery and the farmer's field has been built by human hands. Extension agents, often walking miles through dust and heat, carry the latest knowledge on seeds, pests, and weather patterns to communities that need it most. Yet, as populations grow and climates shift, this human network faces a breaking point. There are simply too many farmers and too few agents to go around. In response, the agricultural world has turned to a powerful new tool: artificial intelligence. Think of this technology not as a robot taking over, but as a tireless digital assistant capable of reading vast amounts of data to offer advice on what to plant, when to water, or how to treat a sick crop. But for this advice to be useful, the farmer must understand it, trust it, and feel comfortable using it. If the computer's reasoning remains a black box, the advice is likely to be ignored.
A team of researchers set out to map the current landscape of this digital revolution in agriculture. They conducted a wide-ranging review of recent studies to see how artificial intelligence is actually being used to support farmers and the agents who guide them. Their work focused on three critical questions: what these systems can do, whether they can explain their reasoning in a way humans understand, and what it takes for farmers to actually trust and adopt them. The researchers gathered and analyzed thirty-two distinct studies published between 2021 and 2026, looking at everything from chatbots that answer farming questions to complex systems that predict crop yields.
The review revealed that the technology is moving quickly from simple information delivery to interactive, personalized advice. In many places, farmers are now using mobile apps and chatbots that act like knowledgeable neighbors, offering real-time guidance on weather, pest control, and crop selection. Some systems have grown sophisticated enough to engage in conversation, answering specific questions in local languages and tailoring advice to the unique conditions of a single farm. For example, one system deployed across four countries has already engaged more than 15,000 farmers, responding to hundreds of thousands of queries. Another project in India uses natural language processing to provide agricultural information in the Odia and Sambalpuri languages, breaking down language barriers that once kept vital information out of reach.
However, the researchers found that building a smart system is only half the battle. The other half is making sure the farmer believes in it. A major theme in the studies was the need for "explainability." This means that when an AI suggests a course of action, it must be able to say why. If a system tells a farmer to apply a specific amount of fertilizer, it should be able to point to the soil data or weather patterns that led to that conclusion. The review noted that while scientists have developed methods to make these computer decisions clearer, there is still a gap between the technical ability to explain a result and the farmer's ability to understand it. Trust is not built on accuracy alone; it is built on transparency. Farmers need to know where the information comes from and feel confident that the advice is safe and fair.
Despite the excitement, the path forward is not without significant hurdles. The review highlighted that technology cannot solve problems that are rooted in infrastructure and society. In many rural areas, the internet connection is too weak to support these advanced tools, and the cost of smartphones remains out of reach for many small-scale farmers. Furthermore, the studies suggested that the most successful systems are those that respect local knowledge and involve farmers in their design. A system that works perfectly in a laboratory may fail in a field if it does not account for local customs, language nuances, or the specific challenges of a region. The researchers found that while many projects focus on how well the technology performs, fewer studies have tracked whether these tools actually lead to long-term improvements in a farmer's income or crop health.
The authors conclude that the future of agricultural advice lies not in replacing the human agent, but in empowering them. The most promising vision is one where artificial intelligence acts as a powerful partner, handling the heavy lifting of data analysis while human agents and farmers provide the context, judgment, and trust that machines cannot replicate. The technology has proven it can generate advice, but the next great challenge is ensuring that advice is understandable, trustworthy, and truly helpful in the real world. Until the digital divide is closed and the systems are designed with the farmer's daily reality in mind, the full potential of this partnership will remain just out of reach.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.