Published on: 2026-05-26
Source: People’s Republic of China in Russian –
An important disclaimer is at the bottom of this article.
Beijing, May 26 /Xinhua/ — China has released its first open large language model (LLM) called “Green Shield,” designed to enhance the effectiveness of plant protection.
The model, developed by Nanjing Agricultural University (NAU) in collaboration with the State Key Laboratory of Agrobiological Safety and more than 30 industry institutes, is designed to provide scientifically grounded agronomic recommendations and control the proper use of pesticides, reported the Science and Technology Daily newspaper on Tuesday.
“China is facing frequent outbreaks of pests in agricultural crops and a problem of pesticide sustainability,” said the project leader, deputy director of the Plant Protection Institute of Nanjing Agricultural University Dun Shamen.
As the publication further quoted, at the presentation on Monday, it noted: “Farmers urgently need professional on-site consultations. However, universal LLMs often provide inaccurate answers to questions about plant protection, and, what is even more important, offer incomplete, sometimes dangerous recommendations regarding the use of pesticides.”
To solve this problem, the team created a specialized text corpus with a volume of more than 2.5 billion tokens, including scientific articles, patents, national standards, and field reports. The corpus covers the main agricultural crops, including rice, wheat, soybeans, vegetables, and fruit trees, and integrates information on pest monitoring, green control methods, and pesticide registration, as stated in the message.
Van Dunbo, professor at the Institute of Information Management at Nanjing Agricultural University, stated that the model is capable of accurately recognizing types of crops, growth stages, and symptoms of plant diseases, and then generating comprehensive growth management strategies.
“Thanks to targeted training, the model works well and recognizes pests with high accuracy,” said Van Dunbo.
Before issuing any recommendation, the model is automatically verified against the national pesticide registration database, checking each chemical substance for compliance with banned lists, authorized crops, and dosage norms. Any non-compliant suggestion is blocked and self-corrected, which prevents improper pesticide use on the crop, Van Dunbo noted.
Van Yuanchao, Vice-Rector of Nanjing Agricultural University, stated that the university will continue field tests and model iterations to create an intelligent tool that will be “understandable, convenient, and effective” for farmers, facilitating the introduction of digital technologies into modern agriculture throughout the entire production chain.
Please note; This information is raw content obtained directly from the source of information. It represents an accurate report of what the source claims, and does not necessarily reflect the position of MIL-OSI or its clients.