People's Daily: The industrialization of artificial intelligence is accelerating

Source: China Academy of Information and Communications Technology, International Data Corporation (IDC) Draftsman: Shen Yiling

Core reading

In recent years, with the continuous improvement of digital infrastructure construction and the acceleration of commercial applications, the development of artificial intelligence industry has entered the "fast lane".

Industrial quality inspection, parts counting, autonomous driving, voice interaction... Artificial intelligence has brought tangible results to enterprises in product research and development, service upgrading, business model innovation, etc., and various artificial intelligence applications have entered production and life and served thousands of households.

The new generation of artificial intelligence is the driving force for the leapfrog development of science and technology, industrial optimization and upgrading, and the overall leap of productivity. At present, the field of artificial intelligence in mainland China presents the development characteristics of two-wheel drive and two-way promotion of technological innovation and industrial application.

Under the joint promotion of both supply and demand, technological innovation achievements began to move from laboratory research to industrial practice on a large scale, and the industrialization process of artificial intelligence accelerated. According to the latest data released by the China Academy of Information and Communications Technology, the scale of the core artificial intelligence industry in mainland China will reach 508 billion yuan in 2022, a year-on-year increase of 18%.

The breadth and depth of applications are expanding faster, and scenario innovation has become a new path

Rabbits, oil droplets, metamorphosis, partridge spots... In the heat of more than 1300 degrees Celsius, Jianzhan shows a brilliant glaze color, each of which is unique. The ancients often used "one color in the kiln, one color out of the kiln" to describe its ever-changing glaze color. In recent years, the domestic construction industry has developed rapidly, but it has also produced chaos such as counterfeit imitation and shoddy charging, and it is difficult to solve the problem through traditional anti-counterfeiting means.

Later, a Jianzhan manufacturer located in Jianyang District, Nanping City, Fujian Province, approached Beijing Megvii Technology Co., Ltd., hoping to use artificial intelligence technology to create an exclusive "electronic ID card" for each Jianzhan technology.

This is not an easy task. On the one hand, it is extremely difficult to distinguish the difference between different Jianzhan and the glaze of Jianzhan will also produce strong reflection, which will bring challenges to artificial intelligence; On the other hand, such projects often require a lot of time to do experimental design.

Faced with these problems, the Megvii team proposed the concept of "algorithm mass production", created an artificial intelligence construction traceability system, and realized the pattern recognition. The team also developed hardware devices with built-in light control to overcome the problem of reflective glazed materials.

"As a result, we have realized the combination of one picture, one light and one code, which not only effectively cracks down on counterfeiting, but also promotes the standardized development of the industry and the upgrading of digital intelligence." Zhou Erjin, head of algorithm mass production at Megvii Research Institute, said.

Like Jianzhen, artificial intelligence applications have evolved in depth, and more and more industries have achieved intelligent upgrades. Industrial quality inspection, parts counting, automatic driving, voice interaction, and even high-altitude parabolic detection, Pu'er tea cake identification, broken skin quick-frozen dumpling detection... Artificial intelligence has brought tangible results to enterprises in terms of R&D speed and process, product service, business model innovation, etc., and "various" artificial intelligence applications have entered production and life and served thousands of households with a wave of digitalization and intelligence.

"AI technology is evolving in the direction of pursuing higher precision, challenging more complex tasks, and expanding the boundaries of capabilities. Scenario innovation has become a new path for AI technology upgrading and industrial growth. Zhao Zhiyun, director of the New Generation Artificial Intelligence Development Research Center of the Ministry of Science and Technology, secretary of the Party Committee and director of the China Institute of Science and Technology Information, believes that the breadth and depth of artificial intelligence applications are constantly expanding, and are penetrating into more industries and more core business fields, promoting the continuous improvement of intelligence in manufacturing, logistics, medical and other aspects, and the importance of intelligent scenarios to economic and social development has gradually become prominent.

New technologies are constantly emerging, and the path of industrialization is gradually clear

With the continuous emergence of new technologies, the path of artificial intelligence industrialization is gradually clear. In recent years, domestic and foreign manufacturers have increased their investment and research and development of huge models, so that the artificial intelligence industry has found a new direction.

New drug research and development benefits from this. HUAWEI CLOUD Pangu Drug Molecule Model is a large model jointly trained by HUAWEI CLOUD and the Shanghai Institute of Materia Medica, Chinese Academy of Sciences, which enables AI-assisted drug design for the whole process of small molecule drugs.

Professor Liu Bing's team from the First Affiliated Hospital of Xi'an Jiaotong University successfully developed a new "super antibacterial drug" using this large model, which is expected to become the world's first new target and new class of antibiotics in nearly 40 years. With the help of large models, the development cycle of lead drugs has been shortened from several years to 1 month, and the development cost has been reduced by 70%.

Rui Yong, chief technology officer of Lenovo Group, believes that the characteristics of large models can be summarized as "one big and three more": "'One big' refers to a large scale of parameters, which is a super-large artificial intelligence model with hundreds of billions of parameters; 'Three more' refers to the use of multi-source, multi-modal, multi-task Internet massive data for training." ”

"On the basis of cloud services, the landing of large models is becoming more and more common. In the past, we often emphasized small models for a certain type of application, such as face recognition, speech recognition, etc. Now with the emergence of large models, users can perform simple processing on pre-trained models to meet their own needs. This can actually help users to better apply AI technology. Zhou Zhengang, vice president of International Data Corporation (IDC) China, said.

Inspur's intelligent customer service robot "Yuanxiaoservice" based on the large model is a typical application of the large model. "Our goal is to combine large models with industry models that fit the needs of the scenario." Chen Bin, director of Inspur Information Service, said.

With strong learning ability, "Yuanxiaofu" can independently learn the knowledge base. At present, it can cover 92% of the consultation questions of end users. For some common technical problems, such as system installation, component abnormalities, etc., the resolution rate reaches 80%.

"The big model is not a showman, it is more derived from the actual needs of the development of the artificial intelligence industry." Liu Jun, vice president of Inspur Information, pointed out that artificial intelligence application scenarios have transitioned from fragmentation to deep integration integration, and in terms of algorithm models, giant models will become the basis for large-scale innovation of artificial intelligence algorithms.

The scale of core industries has grown rapidly, forming a good foundation for industrial development

Artificial intelligence is a strategic technology leading a new round of scientific and technological revolution and industrial transformation. In recent years, the mainland has successively issued a number of policies to encourage the development and innovation of the artificial intelligence industry. Industrial policies such as the New Generation Artificial Intelligence Development Plan, the Notice on Supporting the Construction of a New Generation of Artificial Intelligence Demonstration Application Scenarios, and the Three-Year Action Plan for the Development of New Data Centers (2021-2023) have provided guarantees for the development of the AI industry. Mainland China has also built more than 10 national new-generation artificial intelligence open innovation platforms relying on leading enterprises; Relying on local governments to build a national new-generation artificial intelligence innovation and development pilot zone and a national artificial intelligence innovation and application pilot zone; Promote the large-scale application of artificial intelligence technology in key areas such as medical care and elderly care and urban construction by publishing intelligent product promotion catalogs...

In recent years, the mainland has formed a good foundation for the development of artificial intelligence industry. In addition to the rapid growth of the scale of the core AI industry, the number of artificial intelligence enterprises in mainland China has exceeded 4,200, important breakthroughs have been made in key core technologies such as smart chips and open source frameworks, and the innovation capabilities of iconic products such as smart sensors and intelligent networked vehicles have continued to improve. In terms of intelligent information infrastructure layout construction, in 2022, the total scale of data center racks in use in the country will exceed 6.5 million standard racks, and more than 30 cities are currently building or proposing to build intelligent computing centers.

Experts said that the key to further promoting the development of the artificial intelligence industry, transforming innovation elements into material or knowledge resources, and forming scale effects and scope effects lies in the construction of collaborative innovation platforms. To build an AI collaborative innovation platform, it is necessary to focus on the current application needs, faster realize the business aggregation, resource aggregation and strategic aggregation of AI ecosystem partners, and form an industrial ecology in which the upstream and downstream of the industrial chain cooperate by strengthening the supply of factors at the levels of AI computing power output, service capability optimization and talent training.

People's Daily (Version 05, 25 March 2023)