The automation assembly line of the digital workshop of Fuda Bearing Group Co., Ltd. located in Henghe Town, Cixi City, Zhejiang Province. Photo by Xinhua News Agency reporter Xu Yu

Staff from Delong Software Technology (Tianjin) Co., Ltd. showcase a monitoring software for industrial production.

Photo by Xinhua News Agency reporter Sun Fanyue

  At present, a new round of scientific and technological revolution and industrial transformation are in-depth evolution, artificial intelligence (AI) empowers industrial development in all aspects and in-depth, and industrial intelligence, integration and greenness are accelerating, promoting in-depth adjustments to the global industrial chain and supply chain.

  Change production methods

  my country has rich application scenarios, a super-large-scale market and a huge talent team, which has laid a solid foundation for the development of artificial intelligence and has formed a complete industrial system covering the basic layer, the framework layer, the model layer, and the application layer. The rapid evolution of artificial intelligence technology has also injected new impetus into thousands of industries.

  "Artificial intelligence breaks the boundary between virtual and reality, and promotes fundamental changes in the technological innovation paradigm. At the same time, it breaks through time and space limitations and changes the production and manufacturing methods." Wang Yuxia, chief researcher of digital economy at CEDA Research Institute, gave an example. Before using artificial intelligence technology, the number of stable crystals discovered by humans was 48,000. After introducing the GNoME model of the artificial intelligence material discovery tool, scientific researchers discovered 2.2 million new crystals in a short time, of which 380,000 new crystals have stable structures, becoming the most likely new materials to be synthesized and put into use through experiments.

  Manufacturing is the main battlefield of artificial intelligence. Through large-scale application in the manufacturing industry, artificial intelligence technology and products have also been iteratively upgraded to form two-way empowerment. The Ministry of Industry and Information Technology supports the construction of 11 national pilot zones for artificial intelligence innovation application, carry out a series of in-depth activities for artificial intelligence to empower new industrialization, select 151 typical cases, and set up a number of general big models, industry big models and enterprise benchmarks to provide guidance for the intelligent transformation of the industry.

  The report "Artificial Intelligence Empowers New Industrialization: Paradigm Change and Development Path" released by the Institute of Information Technology and Software Industry of Cydie Research Institute recently believes that it is necessary to gradually realize the intelligent upgrade of the entire manufacturing process around R&D design, medium-sized test and certification, marketing services, operation and management.

  "In the early stage of manufacturing transformation, we can give priority to areas with relatively standardized business scenarios and low technical thresholds to carry out application demonstrations, and use artificial intelligence technology to quickly achieve large-scale deployment; from the medium and long term, we can gradually penetrate the application of artificial intelligence into core links such as design, research and development, production and manufacturing, overcome the problem of integrating artificial intelligence technology with complex industrial scenarios, form a number of practical and feasible intelligent solutions, and ultimately build a full-process, full-factor, and full-scene industry application artificial intelligence system." Wang Yuxia said.

  For example, Changan Automobile has established an intelligent R&D cloud platform for automotive products with three-dimensional digitalization and global collaborative design as the core, and has achieved cross-departmental, cross-enterprise and cross-regional collaboration to ensure design solutions coordination and adaptation. This R&D model can make full use of high-quality resources from all over the world, ensure real-time data transmission and collaboration between research centers in different countries and regions, and improve R&D efficiency and quality.

  Innovative product form

  This year's "Government Work Report" proposes to support the widespread application of large models and vigorously develop new generation smart terminals and intelligent manufacturing equipment such as intelligent connected new energy vehicles, artificial intelligence mobile phones and computers, and intelligent robots.

  "In recent years, artificial intelligence technology represented by big models has accelerated its deep integration with smartphones, connected cars, home appliances, wearable devices, etc., and improved the terminal product usage experience." Wang Yuxia said.

  Enterprises are accelerating the layout of innovative applications of artificial intelligence technology, and the market popularity of smart terminals such as AI mobile phones and AI PCs (personal computers) has gradually increased. International Data Corporation (IDC) predicts that China's smart terminal market shipments will grow by 4% in 2025, with the total shipments of AI personal computers, AI tablets and AI mobile phones increasing by 20% year-on-year.

  Lenovo's latest financial report shows that against the backdrop of the overall decline in the personal computer market, Lenovo's revenue and profit in China both grew by more than 20% year-on-year, and an important growth point is AI PC. Lenovo is the world's first enterprise to deploy and run the DeepSeek big model on the AI ​​PC side. In February this year, Lenovo Tianxi's personal intelligent system deeply integrated the DeepSeek big model, and its capabilities were greatly upgraded.

  "Major manufacturers reshape the human-computer interaction paradigm through technological innovation and scenario implementation, and promote the industry to shift from simple product competition to all-round value creation." Zhao Yan, deputy director of the Institute of Electronic Information of Cydie Research Institute, introduced that smart terminals integrate dedicated AI engines to improve the level of hardware computing power. At the same time, the AI ​​model is embedded into the operating system kernel to create the underlying intelligent center for system operation.

  Not long ago, ZTE released the industry's first full-size small folding phone with DeepSeek embedded in the industry. Users do not need to download it separately, and can directly open the application with one command. In addition, through the Nebula Engine intelligently calls a variety of expert models, users only need to ask specific questions and the mobile phone can automatically select the most suitable "expert" answer.

  "In recent years, we have continuously expanded the application of AI in mobile phones, PCs, tablets and mobile Internet terminal devices to accelerate AI inclusiveness. In the future, we will continue to rely on Nebula's AI core, terminal and cloud integrated technology capabilities, integrate Star Models, DeepSeek, etc., to explore smarter interaction methods and improve user experience." said Ni Fei, senior vice president and president of ZTE's terminal business.

  However, the AI ​​terminal standards have not been established yet, and there is a phenomenon of difficult software and hardware adaptation in the development process, uneven product quality and serious homogeneous competition. Issues such as personal privacy protection, data security and compliance have also attracted much attention.

  Zhao Yan suggested that a new generation of smart terminal industry standards should be jointly built, a general adaptation and application development platform should be created, and the compatibility and interoperability of smart terminals under different operating systems should be ensured, and industrial ecology should be coordinated. The research and development of software and hardware security technologies on both sides of the terminal cloud should be strengthened, laws and regulations should be formulated and issued, and government supervision mechanisms should be improved to effectively solve user data security and privacy issues.

  Accelerate deep fusion

  Zhu Keli, executive director of the China Information Association and founding director of the China Research Institute of New Economics, believes that traditional industrialization relies on factor investment and scale expansion, while intelligent transformation driven by artificial intelligence promotes the leap from production-oriented manufacturing to service-oriented manufacturing through data factor activation, algorithm model iteration and intelligent decision-making optimization. At present, the deep integration of industrial big models and vertical scenes has become a new highlight. The industry model based on the deep learning framework is being implemented at a faster pace, and closed-loop applications from design to production are realized in the fields of electronic information and automobile manufacturing.

  Important progress has been made in empowering new industrialization by artificial intelligence. Zhao Gang, director of Saizhi Industry Research Institute, introduced that the technical foundation is constantly being consolidated, and technologies such as artificial intelligence computing power, intelligent computing operating systems, and industrial big model platforms are accelerating their application in the manufacturing industry. Industrial enterprises have widely accessed models such as DeepSeek, and there are more than 100 industrial big model application platforms. Industrial policies continue to be introduced, and provinces such as Guangdong and Jiangsu have introduced support policies to promote the empowerment of new industrialization of artificial intelligence.

  To promote the deep integration of artificial intelligence and new industrialization, we need to work together in multiple dimensions of technology, ecology and system. Zhukeli analyzed that the technical level should focus on the dual-wheel drive of basic models and industry knowledge to break through the bottleneck of insufficient algorithm generalization capabilities; ecological construction needs to break through data silos and technical barriers; in terms of institutional guarantees, it is necessary to improve the innovation incentive and risk prevention and control system.

  "We must deepen the research and development and application of new technologies such as the Internet of Things, computing power chips, industrial-grade intelligent computing platforms, and industrial big models in the industrial field, break through the technical performance bottlenecks in complex industrial scenarios; strengthen data governance of industrial enterprises, promote industrial data annotation, training and reasoning, and build high-quality industrial data sets; take intelligent manufacturing as the main focus, deepen industrial big models, intelligent bodies and embodied intelligent applications in various industries, and promote the intelligent transformation of the entire chain of manufacturing." Zhao Gang said.

  Zhu Keli said that the fragmentation of industrial data will restrict the effectiveness of model training, and it is necessary to establish a cross-industry data sharing mechanism to promote the circulation of data elements while ensuring security. At the government level, a national-level industrial data trading platform can be built to formulate standards and specifications for data rights confirmation, transactions and use. At the enterprise level, leading enterprises are encouraged to open up intelligent platform interfaces, drive small and medium-sized enterprises to access the ecosystem, and form a development pattern led by chain leaders and coordinated supporting enterprises. (Reporter Li Pengda Huang Xin)

[Editor in charge: Ran Xiaoning]