Analysis of Problems and Countermeasures in the Development of China's Artificial Intelligence Industry

[baoshi.rao]

1. Overview and Characteristics of China's AI Industry Development

1. Wide industrial chain layout with strong specialization

In terms of industrial chain characteristics, within China's AI industry ecosystem, large companies participate extensively based on resource capabilities, covering the foundational, technological, and application layers. China boasts many excellent AI companies, most of which are highly specialized, focusing on technology and application research in specific niche areas. The computer vision field, in particular, has gathered a large number of outstanding startups. However, there are certain differences in the relevance of AI technologies across various application scenarios.

2. Primarily B2B-oriented business

In terms of business models, most companies primarily focus on B2B solutions and services. On one hand, B2B business emphasizes interaction and collaboration with industry clients, which is more conducive to the implementation of AI technologies and products. On the other hand, industry clients have strong demands for improving production efficiency, while the demand for C-end products still needs to be explored. Nevertheless, the C-end product strategies of large companies remain relatively active.

3. High talent costs and significant demand gap

Technologically, machine learning algorithms represented by deep learning constitute fundamental capabilities, but China currently faces relative shortages in talent supply within this field, coupled with weak mobility. This has led to exorbitant costs for high-end research talent, prompting some companies to establish research institutes or laboratories in the United States. This demonstrates that as a typical representative of knowledge-intensive industries, the artificial intelligence sector has significant demand gaps.

4. Strong Support from Traditional Industries and Technologies

On the product front, there remains a lack of truly revolutionary products, with most being improvements to traditional industry products through AI technology. In this process, industries such as healthcare, equipment manufacturing, automotive, and finance have provided substantial support to the AI industry. Through collaborative development and other means, they facilitate the application and commercialization of AI technologies.

II. Problems and Countermeasures in China's AI Industry Development

1. Analysis of Existing Problems

From a business perspective, although some AI companies have applied relatively mature technologies to social life, the corresponding level of commercialization still needs enhancement. For example, Baidu, which previously positioned AI as the new direction for future business, has made some breakthroughs in monetizing AI technologies. However, the full potential of AI-driven performance growth has yet to be realized. Baidu's AI engine, "Baidu Brain," currently supports various business lines. Yet, in the core area of autonomous vehicle development, Baidu has only obtained a testing license for driverless cars in California, USA.

In the past, the slow development of AI was attributed not only to limited computing power but also to insufficient data. Today, almost every activity can be recorded as data if desired. The challenge lies in how to utilize this data to help AI evolve faster and more effectively. Therefore, the first bottleneck in AI commercialization stems from data.

The second bottleneck involves the exploration and construction of more application scenarios. Some AI applications have indeed replaced human labor, with some even surpassing human efficiency. For instance, robots like AlphaGo have demonstrated capabilities comparable to humans in games like Go. However, existing AI applications still fall short of meeting societal needs. In more specialized fields, such as geological exploration and medical diagnosis, human expertise remains essential for evaluation and judgment.

Although the concept of AI has gained public awareness in recent years, widespread application is still far off. Developing mass-market AI products with independent brands—combining software and hardware around human behavior trajectories in wearable, automotive, and smart home scenarios—and achieving scale is currently the most viable path from technology to product and commercialization.

Additionally, the third bottleneck primarily lies in the level of technological R&D. While current AI technological development can meet some commercialization needs, there remains vast room for expansion and deepening.

In terms of AI commercialization applications, especially in the mobile market, the general reality is: for hardware development, more cost-effective chips and system design architectures are needed; for software development, there's a need to explore algorithm training across more domains to uncover demands, along with requiring talent that can integrate software, hardware, and applications in specific algorithm implementations.

2. Countermeasure Analysis

The core strategy should revolve around deep convolutional neural networks, comprehensively developing AI products in computer vision, speech recognition, and natural language processing for large-scale industrial applications. This requires rapid advancements in big data, computing platforms/engines, AI algorithms, and application scenarios, as well as resources, funding, and talent. Methodologically, selecting the most critical vertical industry segments is paramount.

For specific vertical industries, establishing big data centers is essential. This involves implementing big data collection, cleaning, labeling, storage, management, and trading, while building public infrastructure for big data sources and vertical industry knowledge bases. Proprietary big data is the key to success in the AI industry. Chinese enterprises must begin paying special attention to big data collection and utilization. Its importance is comparable to crude oil – multinational corporations regard it as a strategic resource!

Vigorously advance the research and development of AI chips and hardware platforms. This includes FPGA-based deep learning chips, neuromorphic chips and memristive devices, and the establishment of national-level AI supercomputing centers.

Deploy cutting-edge technologies in general artificial intelligence and cognitive intelligence. Strengthen interdisciplinary innovation integrating brain science, cognitive science, and psychology to drive original basic research, supporting China's AI applications and industrial development.

Innovate institutional mechanisms to seize the strategic high ground in AI. Enhance the national scientific and technological innovation system, reform academic and research evaluation systems for input-output. Address major national strategic needs and urgent socioeconomic development requirements through systemic innovation to achieve technological and industrial breakthroughs, bridging all links in "government, industry, academia, research, and application." Examples include establishing a national DARPA (Defense Advanced Research Projects Agency) and China's equivalent of Sandia National Laboratories.