澎湃News Wang Wei Kong Jiaxing
On September 17, the 2018 World Artificial Intelligence Conference opened in Shanghai. The conference “held a new era of artificial intelligence empowerment”, and held nearly 20 theme forums and summits to discuss brain-computer integration, group intelligence, smart chips, and intelligent driving. Participating in the forums and exhibitions are experts and scholars from nearly 40 countries and regions around the world and more than 200 artificial intelligence leaders.
In the latest 2018 China Artificial Intelligence Development Report, Tsinghua University China Science and Technology Policy Research Center revealed the development of artificial intelligence in China from four aspects: technology development and market application: in research, Chinese artificial intelligence in the past 20 years The number of papers and the number of high citations (Note 1) ranked first in the world; in terms of talent investment, Chinese artificial intelligence talents (Note 2) have the second largest in the world, but outstanding talents (Note 3) accounted for less than other leading companies. In terms of industry, the number of artificial intelligence enterprises in China ranks second in the world, and nearly 40% comes from Beijing. In terms of investment and financing, the amount of financing in China's artificial intelligence field accounted for 60% of the world in 2017, which is the most “sucking gold”. ;country.
Global AI paper: China overtakes the United States, winning the championship
According to the report, the output of papers in the field of artificial intelligence in the world entered a growth period in the 1990s, and in 1994, the total number exceeded 20,000. After 2010, the total amount of global artificial intelligence papers declined slightly, and then maintained a steady growth momentum. In recent years, the global artificial intelligence paper has reached the order of more than 100,000 pieces per year, reflecting that this research field is getting more and more attention from researchers.
In the past 20 years (1997-2017, the same below), the output of artificial intelligence papers in China and the United States ranked first and second respectively in the world. China has produced a total of 369,000 papers in the field of artificial intelligence, and its output is 3.8 times that of the third largest in the world.
The global share of papers in the field of artificial intelligence in China increased from 4.26% in 1997 to 27.68% in 2017. In 2006, China's artificial intelligence paper output surpassed the United States for the first time in the world, and has been at the forefront of the world since then.
In addition to China, the results of research in the field of artificial intelligence in India, which is also a developing country, are equally impressive. According to the report, the output of artificial intelligence papers in India grew rapidly in 2011, surpassing the United Kingdom, Germany, France and Canada in 2013, becoming the country with the largest number of papers after China and the United States.
In the past 20 years, from the number of highly cited papers in the field of artificial intelligence, the top three countries are China, the United States and the United Kingdom, and Iran is the eighth in the West Asia. According to the report, the number of highly cited papers in the global artificial intelligence field accounts for 1% of the total number of papers, and the countries that exceed the global average level are Australia, the United Kingdom and China.
From the perspective of the subject classification of artificial intelligence papers, computer science, engineering and automation control systems are the subject of research in the main paper producing countries. In addition to China and India, the number of artificial intelligence papers in the life sciences and biomedical fields of other major paper-producing countries ranks among the top ten in the country.
The research in the field of artificial intelligence is inseparable from the cooperation of researchers all over the world. The above diagram shows a network of cooperation between the 10 countries with the highest number of high-level papers in the field of artificial intelligence. Among them, the larger the circular node, the larger the output of the high-level papers representing the country; the thicker the connections between the nodes, the more the number of collaborative papers representing the state. The report pointed out that China has a large number of high-level papers published in cooperation with the United States, the United Kingdom, and Australia. In addition, some high-level collaborative papers have been produced between the United States, the United Kingdom, and Germany.
70% of the world's AI talents come from universities, and the proportion of outstanding Chinese talents is low.
Through the analysis of AI home institutions and enterprises, the report lists the 20 institutions and enterprises with the largest number of papers in the field of artificial intelligence. Among them, the Chinese Academy of Sciences has published a total of 26,176 papers, ranking first in the world's institutional paper output, and the US IBM company has taken the lead in the corporate world.
Differences in the sources of papers in the field of artificial intelligence in different countries and regions. The output of papers in China, the United States, and France mainly comes from colleges and universities, which represents the academic development of scientific research. The output of papers in Japan, Germany, and the Netherlands is mainly from enterprises, which represents the industry leading the development of scientific research.
Judging from the number of institutions and companies on the list, only one company in China (State Grid) has the number of papers in the field of artificial intelligence that ranks among the top 20 global companies, but six organizations rank among the top 20 institutions. It can be seen that there is still a certain gap between China's academic circles and the industry and the European and American countries.
From a global perspective, international artificial intelligence talents are mainly engaged in universities and research institutions. According to the report, there are 147,914 talents in the world, accounting for 72.3% of the total talents. 31,123 are invested in scientific research institutions such as the National Academy of Sciences and Research Centers, and 6,488 are engaged in commercial enterprises and other for-profit entities.
In terms of investment in artificial intelligence talent, the United States has a total of 28,536 people, accounting for 13.9% of the world total. China ranks second in the world with a total of 18,232 people, accounting for 8.9% of the world total. Most of the outstanding talents of international artificial intelligence are engaged in a few developed countries such as the United States, Britain, Germany, France and Italy. The proportion of outstanding talents in developing countries such as China is obviously low. According to the report, China's proportion of outstanding talents in artificial intelligence is the lowest among the top ten countries in terms of artificial intelligence talents, accounting for only 5.4% of the total number of artificial intelligence talents in the country, while the proportion of outstanding talents in other leading countries is equal. At 11.9% or more.
From the perspective of subdivided research fields, the research areas most favored by international artificial intelligence talents are machine learning, data mining and pattern recognition. Chinese artificial intelligence talents tend to invest in genetic algorithms, neural networks and fault diagnosis.
China has the second largest number of AI companies in the world, and the most popular visual technology
According to the report, as of June 2018, there were 4,025 artificial intelligence enterprises in the world, and the United States ranked first with 2,028, more than twice as many as the second in China. Among the top 20 cities in the world, there are 9 in the United States, 4 in China, 3 in Canada, and 1 in the UK, Germany, France, and Israel. Among them, Beijing has become the city with the largest number of artificial intelligence in the world.
From the technical field, the application technology of Chinese artificial intelligence enterprises is more focused on vision and voice, while the proportion of basic hardware is smaller. From the perspective of the application of artificial intelligence in the industry, vertical applications such as smart driving, drones, AR/VR, big data and data services are the main directions of artificial intelligence enterprises at home and abroad, but they are more important than foreign countries. Domestically, it is 31% higher.
China's AI sector financing accounts for 60% of the world's total
According to the report, since 2013, the scale of investment and financing in the field of artificial intelligence in the world and China has shown a clear upward trend. In 2017, the global investment in artificial intelligence reached US$39.5 billion, and China's total financing reached US$27.71 billion, accounting for 70%. In the global artificial intelligence field, there were 1208 financing events and China accounted for 31%. About 30% of the investment events have received 70% of the global financing, and the overall financing scale of China's artificial intelligence market is in a leading position in the world.
From the perspective of financing rounds, the proportion of early-stage investments by artificial intelligence companies in the seed round, angel round and round A has gradually declined in 2015. However, from the perspective of the investment round of 2017, the A round is still the highest investment round. A large amount of capital flows into the artificial intelligence market, which is dominated by start-ups, which is an opportunity and a challenge for the entire industry.
Note 1: According to the definition of Essential Science Indicators (ESI), highly cited papers refer to the top 1% papers cited in the past decade.
Note 2: Have the research ability and professional technical knowledge of creative labor in the field, and have the output of innovation in the past 10 years.
Note 3: The top 10% of the H factor in the world's artificial intelligence talent dataset