Farmers can obtain and receive customized advice through AI apps on their mobile phones, just like chatting. This is what Parmesh Shah, the global lead of data-driven digital agriculture of World Bank, shared in Wuzhen on Saturday.

Centering on the theme "Digital Agriculture Solutions in the AI Era", the Digital Agriculture Forum, part of the 2025 World Internet Conference Wuzhen Summit on Nov 8, gathered global experts to explore pathways for technological innovation and collaborative ecosystem development to advance global agriculture's digital and sustainable transformation.

AI-powered tailored advice for smallholders

According to data released by the Food and Agriculture Organization of the United Nations in 2021, about one-third of the world's food is produced by smallholder families, while those smallholders are usually troubled by pests and diseases, extreme weather, market fluctuations, and limited access to resources, etc.

In China, according to the third National Agricultural Census in 2016, smallholders accounted for 98.1 percent of all agricultural operating households, constituted 90 percent of the agricultural workforce, and managed over 70 percent of the total cultivated land area, according to the latest data from China's National Bureau of Statistics.

AI gives a chance to fill the digital divide between smallholders and the world's most cutting-edge agricultural technology, and also reduce the cost for them.

Just like the example mentioned above, the app shared by Shah covers 27 kinds of Africa languages and serves over 200 million farmers in the west of Africa, helping smallholders make a 118 percent increase in their revenue.

Another case showed that farmers only need to take a photo of the plant, and through AI analysis, it can help them identify which crops are growing better and which ones are being affected by pests and diseases, Parmesh Shah said. "Previously, it would take days or weeks to diagnose pest and disease issues. However, by the time those days and weeks passed, often the pests had already completely devoured the crops. Even if we had such consultation and answers, they would be useless." Moreover, this app has seen the cost of providing consultation drop from $35 to $0.35 per farmer.

Diverse segments, vibrant fields

In China, various large-model applications are also helping smallholders overcome their production challenges.

Wang Xiaobing, director of information center of the Ministry of Agriculture and Rural Affairs, said the nutrient expert smart fertilization large model, led by the Ministry of Agriculture and Rural Affairs and developed by the Chinese Academy of Agricultural Sciences and the Chinese Academy of Sciences, has established precise models tailored to diverse ecological regions and cultivation practices.

It integrates multi-source, heterogeneous data from satellites, weather stations, and crop growth monitoring. The model has overcome key challenges in generating intelligent, personalized fertilization strategies for both smallholders and large-scale farms, achieving a 30 percent improvement in recommendation accuracy while reducing costs by 90 percent.

Agriculture comprises multiple sectors, including forestry, crop cultivation, and animal husbandry, each with its own distinct branches. This inherent diversity creates fertile ground for applying AI across the agricultural landscape.

Wei Longbao, Qiushi distinguished professor at Zhejiang University and director of the Institute of Food and Agribusiness Management, said in a post-forum interview that many universities and research institutions have already developed various large-scale models for different agricultural domains.

For example, the AI4DLLM, developed with the involvement of China Agricultural University, integrates data from over 2.1 million cows across 40 dairy farming groups in China (2016-24), along with global dairy research materials, covering entire industry chain aspects such as farm management and production decision-making. "This is highly practical," Wei noted. "People can even use voice input—if you can't type, just ask a question, and it replies immediately."

Large models for crop breeding, disaster prevention and mitigation—even custom models created for specific crops like tomatoes and strawberries—have essentially codified agricultural expertise, "This makes it readily accessible and usable, even for those without prior knowledge." Wei added.

China's strength in digital agriculture

Since 2020, China's total grain production has stayed above 660 million tons for five straight years. The output in 2024 broke through the 700 million ton barrier for the first time, reaching a new level of comprehensive production capacity.

At the same time, China is providing a valuable Chinese model in areas such as agricultural technology extension and global poverty reduction.

Shah noted that the World Bank will release a global report on AI-driven agricultural transformation, featuring 126 use cases from around the world. "It will include many compelling cases from China, and it is a valuable contribution of China to the world."

"We can see also number of platforms that help farmers, connect farmers and sell their products directly to market without too many intermediary," said Dejan Jakovljevic, chief information officer of the UN Food and Agriculture Organization and director of the Digital FAO and Agroinformatics Division. "China's experience in applying digital technologies, from precision farming and smart logistics to e-commerce and rural revitalization, serves as a valuable reference for developing countries."

Song Yi contributed to this story.