Current Situation and Future of Artificial Intelligence in China

Although "artificial intelligence" (AI) has become a well-known concept, there is no general consensus on the definition of AI. The traditional idea of AI development is to study how human beings produce intelligence, and then let machines learn the way of thinking and behavior of human beings. John McCarthy, the proponent of the concept of modern artificial intelligence, argues that machines do not necessarily need to think like human beings to acquire intelligence. The key is to enable machines to solve problems that the human brain can solve.

The fourth industrial revolution is approaching, and artificial intelligence has gradually stepped into reality from science fiction. Since the concept of artificial intelligence was first put forward in 1956, the development of artificial intelligence has experienced several ups and downs. With the breakthrough of core algorithms, the rapid improvement of computing power and the support of massive Internet data, AI has finally ushered in a qualitative leap in the second decade of the 21st century and become the focus of global attention. Since AIphaGo defeated Li Shishi in 2016, the excitement and worry about the development of AI are inextricably intertwined.

Even so, countries around the world have realized that AI is the key arena for competition between countries in the future, and have begun to deploy AI development strategies in order to occupy the historical highs of a new round of scientific and technological revolution. For China, the development of AI is a historic strategic opportunity, which is crucial to alleviate the pressures of future population aging, cope with the challenges of sustainable development and promote the transformation and upgrading of economic structure.

This paper describes the development of AI in China from the perspectives of scientific and technological output, talent input, industrial development and market application, development strategy and policy environment.

Science and Technology Output and Talent Input

1. Paper Output: The total number of Artificial Intelligence Papers and highly cited papers in China are the first in the world. China's global share of Artificial Intelligence papers increased from 4.26% in 1997 to 27.68% in 2017, which is far ahead of other countries. Universities are the absolute main force in the output of Artificial Intelligence papers. Among the top 100 institutions in the world, 87 are universities. Artificial Intelligence Thesis Output of China's top universities is outstanding all over the world. Moreover, China's highly cited papers show a rapid growth trend, and surpass the United States to become the world's number one in 2013. But in the global ranking of enterprise papers output, only the State Grid Corporation of China ranks among the top 20 in the world. From the perspective of discipline distribution, computer science, engineering and automatic control system are the most widely distributed disciplines of Artificial Intelligence papers. The impact of international cooperation on the output of AI papers is obvious. In high-level papers, 42.64% of China's papers are published through international cooperation.

2. Patent application: The number of patents in China is slightly ahead of that in the United States and Japan, and the performance of the State Grid is outstanding. China has become the world's largest country in the distribution of AI patents, slightly ahead of the United States and Japan, and China, the United States and Japan account for 74% of the total number of patent publications in the world. Global patent applications mainly focus on speech recognition, image recognition, robotics and machine learning. Among the top 30 institutions holding AI patents in China, scientific research institutes and universities and enterprises perform equally well, with the proportion of technological inventions accounting for 52% and 48% respectively. The main patentees in enterprises show great differences. Especially in the past five years, the technology of AI in China's State Grid has developed rapidly. The amount of patented technology in domestic distribution is much higher than that of other patentees, and ranks fourth in the global ranking of enterprises. Patent technology in China is concentrated in the fields of data processing system and digital information transmission, among which the related patents of image processing and analysis account for 16% of the total number of inventions. Electric power engineering has also become an important field in the distribution of artificial intelligence patents in China.

3. Talent input: The total amount of AI talents in China ranks second in the world, but the proportion of outstanding talents is low. By 2017, China had 18,232 AI talents, accounting for 8.9% of the world total, second only to the United States (13.9%). Universities and scientific research institutions are the main carriers of AI talents. Tsinghua University and the Chinese Academy of Sciences have become the institutions with the largest input of AI talents in the world. However, according to the high H factor (also known as H index, used to evaluate the scientific performance of scientists), only 977 outstanding Chinese talents, less than one fifth of the United States, ranked sixth in the world. Enterprise talent investment is relatively small, high-intensity talent investment enterprises are concentrated in the United States, and only one Huawei enterprise in China has entered the top 20 in the world. Artificial intelligence talents are concentrated in the East and central part of China, but some western cities such as Xi'an and Chengdu are also outstanding. International AI talents are concentrated in machine learning, data mining and pattern recognition, while the research field of AI talents in China is relatively scattered.

Industrial Development and Market Application

1. Enterprise Scale: The number of AI enterprises in China is the second largest in the world, and Beijing is the most concentrated city of AI enterprises in the world. As of June 2018, a total of 4925 AI enterprises were monitored in the world, of which 2028 were American AI enterprises, ranking first in the world. China (excluding Hong Kong, Macao and Taiwan) has 1011 AI enterprises, ranking second in the world, followed by Britain, Canada and India (Figure 1).

In terms of city scale (Figure 2), among the top 20 cities in the number of AI enterprises in the world, 9 are in the United States, 4 are in China, 3 are in Canada, and 1 is in Britain, Germany, France and Israel. Among them, Beijing has become the city with the largest number of AI enterprises in the world, followed by the old one.

Please read the Chinese version for details.