Recently, the China Artificial Intelligence Industry Development Alliance was established. According to reports, the alliance will rapidly promote the application of artificial intelligence technology in manufacturing, health care, life services, urban governance and other scenarios to enhance industrial development capabilities and levels. At the same time, it will integrate the resources of the entire industry chain and promote the accumulation and transformation of artificial intelligence technology achievements and resources. The industry's relatively unanimous view is that the artificial intelligence industry chain can be divided into the base layer, technology layer and application layer. In response to this, the global giants have spread their core technologies and actively deployed them. At the same time, the artificial intelligence industry chain in China has also been "exposed." More than 20 A-share companies are "occupying the beach" industry chain. Foundation layer: Chinese chip tries to "curve overtaking" The base layer mainly includes computing chips, big data, and storage. According to Microsoft’s latest research report, in the field of artificial intelligence, the traditional chip computing architecture cannot support the demand for large-scale parallel computing such as deep learning. This requires new underlying hardware to better reserve data and speed up the calculation process. The base layer mainly uses hardware as the core, including hardware, neural network chips, sensors, and middleware for performance acceleration such as GPU/FPGA, which are prerequisites for supporting artificial intelligence applications. These hardwares provide computing power for the entire artificial intelligence operation. Currently, most of them are traditional international IT giants. At present, in the GPU field, Nvidia is leading the industrial-grade ultra-large-scale deep-network acceleration, and has launched the Tesla based on Volta, the first processor with a speed exceeding 100TFlops. Intel is mainly focusing on the FPGA-building industry and has introduced artificial intelligence chips that mimic the human brain. Google also introduced a second-generation TPU chip to provide chip support for its open source TensorFlow framework. In addition, there are many start-up companies in this field, such as the China Star Micro, Cambrian and Xijing Science and Technology. However, in terms of industrial layout capabilities and R&D capabilities, they cannot match the aforementioned giants. The "Economic Information Daily" reporter noted that China's exploration of human brain chips is not too late. In 2015, led by the School of Computer Science and Technology of Zhejiang University, Zhejiang University and Hangzhou University of Electronic Science and Technology collaborated to develop the first brain-based chip based on silicon material (CMOS) supporting pulsed neural network (SNN) - "Darwin" chip. . The research group believes that although Darwin is the first brain-like chip to support pulsed neural networks in China, it still has a certain distance from the international advanced level of the IBM TrueNorth chip. China's exploration of human-like brain computing is currently more concerned with the "Cambrian" task force led by Chen Yunzhao and Chen Tianshi of the Institute of Computing Technology, Chinese Academy of Sciences. The relevant experts of the Institute told the reporter that the energy efficiency of the Cambrian deep learning processor has been improved by two orders of magnitude compared with mainstream CPUs and GPUs, and it has a strong market competitive advantage. Since 2017, the "Cambrian" research group has obtained a special fund of 18 million yuan from the Chinese Academy of Sciences for 18 months for project research and development and industrialization. According to the Research Center of Intelligent Processors of the Institute of Computing Technology, Chinese Academy of Sciences, this special fund of 10 million yuan is used for basic research of artificial intelligence chips to explore the architecture and algorithms of the next generation of artificial intelligence chips and in some new scenarios (such as AR Application Development Method in /VR). This will lay the scientific and technological foundation for China's participation in the international chip market in the smart era. Special funds are used in the promotion of Cambrian chips in various smart cloud servers, smart terminals, and smart robot markets on the other. They strive to establish the position of the Cambrian chips in the smart chip market within 18 months. According to the person in charge of the Institute of Computing at the Chinese Academy of Sciences, "'London' is a road, the Cambrian is a road, and cooperation with IBM, Intel, and AMD is another way. 'Godson' is a 'people have me' road The Cambrian walked on the road of 'turning overtaking'." Technical layer: breakthrough in domestic speech recognition algorithms The technology layer includes an algorithm platform, image recognition, natural language recognition processing, and intelligent robots. At present, the domestic artificial intelligence technology platform focuses mainly on the fields of computer vision, speech recognition and language technology processing. The development momentum of domestic technology layer companies is also swift and violent. Among them, representative enterprises include HKUST, Feilin, and Hitachi. Tonghuasheng (Lingyun), Horizon, SenseTime, Yonghong Science and Technology, Science and Technology, and Cloud Knowledge. A research report by McKinsey comprehensively and meticulously reviewed the development status of artificial intelligence in China. McKinsey believes that China is equivalent to other countries in algorithm development. Chinese researchers have made breakthroughs in developing algorithms for speech recognition and targeted advertising. Thanks to the global open source platform, Chinese companies can quickly replicate the most advanced algorithms developed elsewhere. Experts from the China Academy of Science and Technology Strategy Institute told reporters that “the development of the artificial intelligence industry is inseparable from the support of massive data, and the amount of data training affects the maturity of the algorithm.†Ari Yun, Director of iDST Chu Min, said, algorithm, data , computing platforms, users, business models, using the Internet's thinking to string together these five factors, artificial intelligence can be very fast iteration. Using the feedback data to update the model faster, the effect will be better after such a positive cycle. Even if it is the same algorithm, as long as the continuous feedback data and continuous optimization, after a month or two, its ability will be much better. Domestic Shuguang Company has established a space star chart, Zhongke Sanqing, and Shuguang Yitong with a large number of companies to lock in the data. The Space Star map focuses on the geospatial data processing and visualization applications. Sinochem Sanqing is a joint venture between Shuguang and the Institute of Physics of the Chinese Academy of Sciences. It focuses on the forecasting, warning, treatment assessment, and emergency response of atmospheric, water, and soil pollution. In addition to traditional IT companies grabbing data resources, in fact, China has also emerged many companies that operate and operate data, such as data halls, star map data, and percentage points, and more public data open platforms have emerged. Data shows that in 2016, the total amount of data in China accounted for 14% of the total global data. It is predicted that by 2020, China's total data will account for 20% of the total global data. By then, China will become the world’s largest data resource country and global data center. The Innovation Workplace Artificial Intelligence Strategy White Paper shows that data privacy and data security put forward policies, regulations, and regulatory requirements for the establishment of cross-industry, cross-domain big data models for artificial intelligence technologies. The practitioners of vertical fields set out from the commercial interests, and also define the basic rules and boundaries for the sharing and circulation of data. In addition, the accumulation of data in many traditional industries in terms of the degree of standardization and circulation efficiency is still far from reaching the full potential of artificial intelligence technology. McKinsey said that China's big technology companies collect data through their proprietary platforms, but China is lagging behind the United States in creating a data-friendly ecosystem, lacking uniform standards and cross-platform sharing. From the perspective of countries in the world, opening up government data helps the private sector to innovate, but the public sector in China is relatively open to the public. Application layer: driving, medical and other hot topics The application layer includes driverless, smart security and smart medical. From a global perspective, IBM first deployed artificial intelligence applications, "universal Watson" to promote multi-industry changes. Baidu launched the "Baidu Brain" program, focusing on the layout of driverless cars. Google's artificial intelligence business is more complicated, and many areas are blooming, including AlphaGo, driverless cars, and intelligent surgical robots. Microsoft stays ahead of the world in language semantics recognition and computer vision. According to Liu Qingfeng, Chairman of HKUST News and Communications, 2017 is the year that China's artificial intelligence applications have landed, becoming a watershed for the development of artificial intelligence industry. He believes that the application is the absolute principle of the development of artificial intelligence, and only technology can be applied in various fields in order to develop. Ma Huateng, chairman of Tencent Group, said that there is no scenario to support artificial intelligence research in the air. Over the years, the rapid development of artificial intelligence technology has allowed artificial intelligence applications in personal assistants, the automotive sector, healthcare, security, e-commerce retail, finance, and education to cover all aspects of life. Zhang Yaqin, president of Baidu, stated that Baidu needs to establish an ecosystem in the artificial intelligence era. There is no use of artificial intelligence in the scene. Baidu's strategy for the next 10 to 20 years is betting on the artificial intelligence field, and all of the company's resources and technology are tilted toward it. Artificial intelligence has a broad application prospect in the automotive field, of which autopilot is most concerned. In the field of automated driving, many manufacturers have been deep-growing for several years, which makes 2016 a fully competitive year for autonomous driving. This year, Baidu Smart Cars officially unveiled and demonstrated to the world Baidu's leading technologies in the fields of high-precision map manufacturing and auto-driving environment perception, and released an auto-pilot open platform. Through the application of artificial intelligence technology, the safety and efficiency of public transportation systems can be improved. Automated driving vehicles can also reduce traffic accidents, relieve traffic pressures, and play an important role in realizing command traffic. Alibaba cooperated with the Hangzhou Municipal Government to make the urban traffic more intelligent and reduce congestion by integrating traffic signal lights of artificial intelligence technology, and increased traffic flow by 11% in certain areas. Geely Automobile set up a new generation of core business systems as a whole to achieve the online and data-based operations of traditional businesses, helping Geely Automotive lead the "Internet +" trend of the automotive industry. Recently, medical robots developed by relevant teams of the National University of Defense Technology were announced. The robot can provide integrated intelligent medical services such as registration, medical treatment, and physical examination through the use of the supercomputer's big data calculation and artificial intelligence technology, including three major functional systems: smart registration, intelligent diagnosis and treatment, and intelligent and healthy physical examination. Baidu conducts research and development in four areas: medical O2O intelligent triage, intelligent intelligence, precision analysis and precision medicine, and new drug research and development based on big data. It expects to use the computing power and the most advanced algorithms of hundreds of thousands of servers. Apply to medical and health fields. In the past two years, home appliance companies such as Changhong, Midea, Gree, and Galanz have been transforming into smart manufacturing, and have been laying out robot manufacturing and application areas. At the same time, almost all home appliance manufacturers are based on "SmartHome", which integrates artificial intelligence and smart home more closely. However, the experts interviewed said that there is still a long way to go between the business needs of most traditional industries and the cutting-edge scientific and technological achievements of artificial intelligence. The combination of mobile Internet applications and artificial intelligence technologies for general consumers is still in the exploratory stage.