然而，图像识别软件并非百分之百准确，扫描后的内容仍需人工校对与甄别。在这一过程中，团队联合 中国农科院等机构统一数据格式和指标体系，通过建立"数据清洗—标注—审核—入库"的全流程质控体 系，剔除重复、低质数据。"就像教育需要优质教材，训练大模型同样离不开高质量的农业数据。"王耀 君解释说，"这是决定神农大模型能不能立得住的关键一步。" 仅有书本知识，培养出的大模型多半只能"纸上谈兵"。为了让神农大模型真正"接地气"，团队来到全国 20多个省份，收集真实的土壤成分、灌溉记录、病虫害记录、气象灾害影响等全链条数据。这些带着泥 土气息的一手资料，与书本理论相互校验、融合，共同构成了大模型理解真实世界的基础。 算力则是另一道难题。通常情况下，提高算力意味着采购昂贵的芯片，这对本就没有多少经费来源的团 队来说难以承受。"我们不能走堆砌算力的一般路线，必须用更聪明的算法来实现目标。"王耀君告诉记 者，团队创新性地采用了MOE架构（混合专家模型），结合模型压缩与剪枝算法，在有限条件下优化 训练效率，显著降低大模型训练和推理的算力成本。 一走进中国农业大学信息与电气工程学院副教授王耀君的办公室，记者便被一台通体白色的设备所吸 ...

Core Viewpoint - The article emphasizes the growing importance of agriculture in discussions at the World Economic Forum, particularly in the context of AI as a key driver for productivity and resilience in food systems amid global economic and environmental challenges [3][4]. Group 1: AI in Agriculture - AI in agriculture is not hindered by technology but is approached with caution due to the complexity and sensitivity of real-world production systems [4]. - Unlike finance or internet sectors, agriculture lacks scalable applications despite having numerous concept validation projects. The challenges vary significantly between developed and emerging markets, with data fragmentation and infrastructure costs being major issues in developed regions, while usability for smallholders is critical in emerging economies [5]. - The low tolerance for error in agricultural technology adoption leads to a slower acceptance of new technologies compared to other industries, making caution a norm in the expansion of agricultural AI [5]. Group 2: Shift from Yield to Resilience - The focus of agricultural AI is shifting from merely increasing production to enhancing system resilience, as agriculture contributes significantly to greenhouse gas emissions and environmental degradation [7]. - Advanced data analysis and decision support technologies are beginning to reconcile the trade-off between increasing yields and reducing environmental impact, moving from a binary choice to a more manageable range of options [7]. - The discussion around food security is evolving from simply having food available to ensuring stability in food supply amidst various global risks [9]. Group 3: China's Role in Agricultural AI - China is viewed as a significant case study for agricultural AI practices, with a focus on systemic thinking that integrates technology, breeding, chemicals, machinery, and data into a cohesive production logic [11]. - The Chinese approach emphasizes practical applications of AI in specific scenarios like pest identification and weather risk assessment, making it more relevant to farmers' daily decisions [11]. - China's advancements in agricultural digitalization provide a practical testing ground for AI, with improved infrastructure and data accessibility facilitating the transition from demonstration projects to everyday decision-making [11].

谢赫认为，中国经验最值得关注的，并非某一种模型或单一产品，而是一种系统性思维。 在近几年的达沃斯论坛上，围绕人工智能（AI）的讨论几乎无处不在。从金融、制造到能源与医疗， AI被反复描绘为下一轮生产力跃迁的核心引擎。但在今年世界经济论坛年会的讨论中，一个并不总是 站在聚光灯下的领域，正在悄然成为多场讨论的"底层议题"——农业。 在全球经济增长动能放缓、气候风险上升和地缘政治不确定性加剧的背景下，粮食系统被视作关乎稳定 与安全的基础变量。本届达沃斯，围绕"负责任地推动创新规模化应用""在地球限度内实现增长"的讨 论，几乎都会回到一个现实问题：在土地、水资源和环境约束不断收紧的情况下，如何维持粮食供给、 稳定价格，并增强系统韧性。 AI为何在农业领域"走得慢" 在达沃斯接受第一财经专访时，先正达集团首席信息与数字官费罗兹·谢赫（Feroz Sheikh）形容，农业 中的AI并不是"技术跟不上"，而是"被迫更加谨慎"。在他看来，农业AI的核心难题，并不在算法或算 力本身，而在于它必须嵌入一个高度复杂、风险极其敏感的真实生产体系。 从"增产"到"韧性"的逻辑转变 在世界经济论坛的语境中，农业AI的重要性，并不只体现在 ...

Core Insights - Nanjing Agricultural University has developed an open-source vertical large language model named "Si Nong" aimed at the general agricultural sector [1] - The model incorporates data from various sub-disciplines including animal science, agricultural economics, agricultural resources and environment, horticulture, smart agriculture, veterinary medicine, plant protection, and crop breeding, creating a comprehensive agricultural dataset [1] - "Si Nong" has been released in two parameter sizes: 8 billion and 32 billion, with an open-source strategy intended to lower the barriers for AI applications in agriculture [1] Data Collection - The dataset for "Si Nong" includes nearly 9,000 books, over 240,000 academic papers, and nearly 20,000 policies and standards, ensuring a robust foundation for agricultural research [1] Application and Innovation - The open-source approach is designed to facilitate secondary development and innovative applications by research institutions, enterprises, and developers, fostering an ecosystem for smart agriculture applications [1]

Core Viewpoint - The "Shanxi Province Smart Agriculture Action Plan (2026-2030)" has been officially released, focusing on the integration of digital technology with agriculture to promote high-quality development in modern agriculture in Shanxi [1] Group 1: Development Goals - By the end of 2026, the agricultural production informationization rate is expected to reach over 30%, establishing initial smart agriculture public service capabilities [1] - By the end of 2028, the provincial agricultural and rural big data platform will be basically completed, with the informationization rate increasing to over 32% [1] - By the end of 2030, the aim is to achieve widespread promotion of smart equipment and technology models, with a target informationization rate of around 35% in key areas [1] Group 2: Key Tasks - The plan includes three main tasks, focusing on enhancing public service capabilities by building an agricultural and rural big data platform and developing smart agriculture models for breeding and disaster warning [1] - The plan aims to create a unified agricultural land map and establish digital archives for arable land, along with an integrated observation system [1] Group 3: Application Expansion - The action plan covers key industries such as grain and economic crops, livestock, and fisheries, promoting technologies like "Beidou+" precision planting and intelligent water and fertilizer integration in grain production [2] - In economic crops, the plan promotes smart management in vegetable and fruit orchards, while in livestock, it focuses on electronic records and precision environmental control [2] - The fisheries sector will emphasize factory-based recirculating aquaculture systems [2] Group 4: Implementation Support - To ensure the plan's implementation, a provincial-level work promotion mechanism will be established, along with increased financial support [2] - The plan aims to enhance the capabilities of grassroots agricultural technicians and new business entities in smart agriculture applications [2]

Core Points - The project for improving corn yield in 2025 has been launched in the Togtoh County of Hohhot, Inner Mongolia, with a contract amount of 12.588 million yuan awarded to Xiaocao Digital Agriculture Ecological Co., Ltd [1] - Xiaocao Digital will implement a smart water and fertilizer integration system, contributing to the modernization of regional agriculture [1][2] - The project aims to optimize agricultural structure and upgrade the industry in Togtoh County, providing a model for Xiaocao Digital's business replication and regional promotion in smart agriculture [2] Company Summary - Xiaocao Digital has been recognized for its technical strength and market acceptance in the smart agriculture sector, which supports its future performance growth and potential transition to the Beijing Stock Exchange [1] - The company is leveraging "Agricultural AI + Smart Hardware" to create a closed-loop management model for precise resource allocation and intelligent crop management [2] - The project aligns with national food security strategies and the acceleration of agricultural modernization, positioning Xiaocao Digital to expand its order scale and enhance profitability [2]

Core Insights - Zhongnong Bodin Intelligent Agricultural Equipment Co., Ltd. made a significant debut at the China International Agricultural Machinery Expo, showcasing a comprehensive range of products and securing a large exhibition space [1][2] - The company is a subsidiary of Bodin Precision Technology Group and has developed a full lifecycle business model that integrates technology research, core component manufacturing, and complete machine assembly [2][3] - The agricultural machinery industry in China is characterized by a fragmented market with over 8,000 companies, leading to a significant technological gap compared to international competitors [3][4] - The introduction of intelligent tractors by Zhongnong Bodin aligns with national policies aimed at upgrading agricultural machinery, marking a pivotal moment for the industry [3][5] Company Overview - Zhongnong Bodin's tractors are primarily self-developed, with approximately 95% of the entire machine and 100% of core components produced in-house, making it unique in the global agricultural machinery sector [1][2] - The company has successfully launched an automated production line capable of producing 50,000 tractors annually, reflecting its commitment to innovation and efficiency [3][4] - The integration of core technologies allows Zhongnong Bodin to enhance performance while significantly reducing costs, with reported efficiency improvements of over 25% and energy savings of more than 10% compared to similar models [5][6] Market Dynamics - The Chinese agricultural machinery market is expected to undergo transformation due to the introduction of intelligent machinery, with Zhongnong Bodin positioned to lead this change [3][6] - The company has received substantial interest from dealers, with over 4,200 tractors ordered during the expo, indicating strong market demand [6] - By 2026, Zhongnong Bodin plans to reach an annual production capacity of 50,000 tractors and will also sell 50,000 powertrain systems to enhance the competitiveness of smaller enterprises in the industry [6]

Group 1: Overall Financial Performance - In the first three quarters of 2025, the company's total revenue reached 20.744 billion CNY, a year-on-year increase of 3% [3] - The net profit attributable to shareholders was 2.57 billion CNY, reflecting a significant year-on-year growth of 92.56% [3] - In Q3 2025, the company achieved a revenue of approximately 7.184 billion CNY, marking a year-on-year increase of 1.94% [3] Group 2: Revenue Breakdown by Business Segment - Feed products generated revenue of 13.423 billion CNY, a decline of 4.67%, accounting for 64.71% of total revenue [3] - The pig farming segment reported revenue of 4.903 billion CNY, a year-on-year increase of 49.03% [3] - Seed products achieved revenue of 513 million CNY, showing a remarkable growth of 81.07% [3] Group 3: Profitability Analysis - The feed segment contributed a profit of approximately 400 million CNY, while the pig farming segment also reported a profit of around 200 million CNY [3] - The seed segment incurred a loss of approximately 15 million CNY [3] - Total losses from other businesses and public expenses amounted to about 480 million CNY [3] Group 4: Cost Management - Selling expenses decreased by 3.58% to 817 million CNY, while management expenses fell by 5.66% to 1.019 billion CNY [3] - Financial expenses were reduced by 9.51% to 359 million CNY [3] Group 5: Business Segment Performance - In the feed business, external sales volume reached 4.0162 million tons, a year-on-year increase of 2.9% [4] - The pig farming segment saw an output of 3.1608 million pigs, generating revenue of 4.903 billion CNY [5] - The seed segment sold 20.44 million kilograms of crops, a significant increase of 151% year-on-year, with sales revenue exceeding 500 million CNY [5] Group 6: Seed Business Developments - The company ranked third among the top 20 enterprises in national seed sales, leading in the private seed industry [6] - The company has developed six nationally approved high-yield soybean varieties, contributing to domestic self-sufficiency [7] - The company holds a leading position in the domestic market for genetically modified traits, with 59% of approved corn varieties utilizing its traits [8]

Core Insights - The 2025 World Agricultural Technology Innovation Conference (WAFI) opened in Beijing, focusing on "Practicing the Big Food Concept and Building a Resilient Food Supply System" with nearly 800 participants from around 100 countries and regions discussing global agricultural technology issues [1][2] Group 1: Conference Overview - The conference has gained significant influence since its inception in 2023, serving as a high-end platform for international cooperation and innovation in agricultural technology [2] - The event is co-hosted by several prominent organizations, featuring a comprehensive framework that includes an opening ceremony, seven thematic meetings, a World Agricultural Technology Expo, and over 40 parallel meetings and international exchange activities [2] - The expo covers approximately 10,000 square meters with over 150 exhibitors, showcasing various national and international agricultural innovations [2] Group 2: Global Cooperation and Challenges - Kenneth Quinn, Honorary Chairman of the World Food Prize Foundation, emphasized Asia and China as key players in addressing global food security challenges, projecting a global population of 10 billion by 2049 [3] - Experts from various countries expressed a desire to deepen agricultural cooperation with China, highlighting the value of the WAFI as an international exchange platform [3][4] Group 3: Technological Advancements - Enhancing agricultural productivity is crucial for building a resilient food supply system, with calls for increased international agricultural technology cooperation and investment in developing countries [5] - The conference highlighted the importance of modern technologies such as genome editing, AI, and machine learning in addressing global food system challenges [5][6] - The release of the Shennong Model 3.0 by China Agricultural University marks a significant advancement in agricultural AI, providing comprehensive agricultural knowledge and decision-making capabilities [7]