AI can transform China's factories, if data wakes up
Editor's note: China is accelerating efforts to upgrade its manufacturing sector with artificial intelligence technology. Lu Chuncong, head of the China Academy of Industrial Internet, spoke to 21st Century Business Herald about what companies should do to catch up with the trend. Below are excerpts of the interview. The views don't necessarily represent those of China Daily.
China has more than six million manufacturing enterprises and boasts a complete and comprehensive industrial system. This creates a perfect scenario for the application of AI technology and presents vast opportunities for digital and intelligent transformation.
Traditional manufacturing, which accounts for about 80 percent of the country's manufacturing industry, has an urgent need to reduce costs, improve quality and efficiency, pursue eco-friendly development and ensure operational safety.
Reforming traditional manufacturing with AI technology will provide valuable real-world data and complex scenarios that the technology needs, and drive the further development of AI chips, industrial large models, intelligent sensors and other key industries.
Currently, the biggest challenge to the use of AI in the manufacturing sector is that although the sector generates an immeasurable "data torrent" every day, a vast amount of the data remains "dormant".
This is partly due to the different formats and standards the data come in, which make it difficult for AI to interpret them in a unified manner. Many data are scattered across different equipment, production lines, workshops and even management systems, forming data silos that are hard to integrate. In addition, the causal relationships and business logic underlying the data are often complex.
This leads to a dilemma where data, though abundant, are of low-quality, thus preventing their effective utilization. To address this challenge, a shift from physical aggregation to semantic interoperability is needed, so that AI will not only "see" the data but also "understand" the business logic behind them.
Some enterprises are hesitant to use AI technology mainly because they are worried they may not be able to transform their methods of production, and that the transformation may be too costly and even pose security risks.
Such a transformation should never follow a one-size-fits-all approach. Major enterprises should take the lead in fully advancing the transformation by applying AI technology in more scenarios and leading efforts to build high-quality datasets for better collaboration among various companies.
For the vast number of small, medium and micro-sized enterprises with limited capital and technological strengths, the strategy should be "small-scale, fast, lightweight and targeted".
These companies are recommended to use relevant national big data platforms to deploy lightweight edge AI agents to address their most pressing needs.
When choosing among different AI technologies and evaluating their input-output performance, companies should focus on solving their actual business needs and avoid underutilization of the AI technology they have decided to deploy.
Security is a priority, as many enterprises worry about data breaches. They should adopt technologies such as edge computing and federated learning to protect trade secrets.
A long-term view is required when companies evaluate the returns on the investments they make to deploy AI. They should consider not only short-term hardware and software procurement costs, but also the improvement in product quality, enhanced supply chain resilience, reduction in energy consumption and shorter R&D cycles produced by the use of AI.
































