工业智能体“进车间”：中国制造业把AI用在“刀刃”上

Core Viewpoint - The Chinese manufacturing industry is shifting its focus on artificial intelligence (AI) applications from general office scenarios to frontline production after two years of experimental deployment. Midea Group's digital subsidiary, Meiyun Zhishu, has launched an industrial intelligence matrix and the Meiqing AIGC 3.1 platform, aiming to achieve approximately 900 million yuan in cost savings through AI applications by 2026 [1][2][8]. Group 1: AI Implementation in Manufacturing - Midea has deployed over 40 industrial intelligent agents selected from thousands within the company, achieving large-scale applications in 158 core business scenarios, covering R&D, manufacturing, supply chain, quality, logistics, finance, human resources, and overseas services [1][6]. - In production management, Midea utilizes intelligent agents for loss analysis, enabling quick identification of issues causing production losses, which previously required manual checks every few hours [6]. - In the supply chain, intelligent agents can quickly assess the impact of disruptions, such as typhoons, and provide alternative supply solutions [7]. Group 2: Cost Savings and Efficiency - Midea's AI applications are projected to save approximately 40 million yuan in 2023, 180 million yuan in 2024, 600 million yuan in 2025, and reach 900 million yuan in 2026 [8]. - In human resources and legal fields, AI applications can reduce costs by up to 90% in processes like resume screening and contract review [8]. Group 3: Industry Trends and Policy Support - Other manufacturing companies, such as Haier and Skyworth, are also advancing similar AI applications, with Haier's industrial model already implemented in over 40 scenarios across various sectors, achieving a 10% reduction in labor and energy costs in some processes [9]. - The Chinese government is promoting AI in manufacturing through initiatives aimed at deepening the application of 3-5 general models in the industry by 2027 and cultivating 1,000 high-level industrial intelligent agents [9]. Group 4: Challenges in AI Adoption - The promotion of industrial intelligent agents faces challenges, particularly in small and medium-sized enterprises, due to reliance on business participation and data quality, high costs and recruitment difficulties for AI talent, and unclear application scenarios [10][11]. - Companies are advised to first address their digitalization shortcomings before scaling AI applications [11]. Group 5: Future Directions - The future direction for AI in manufacturing is the integration of intelligent agents into traditional industrial software rather than existing as standalone tools, with AI providing decision support in areas like sales forecasting and production planning [12].