An article published in Nature makes a detailed analysis of the current AI development in China from the aspects of paper influence, core applications, hardware and talents.

In 20 17, China formulated the "development plan for a new generation of artificial intelligence", which described the grand blueprint for the development of artificial intelligence in China in the next decade and established the "three-step" goal:

By 2020, the overall technology and application of artificial intelligence will be synchronized with the world's advanced level;

By 2025, the basic theory of artificial intelligence will achieve a major breakthrough, and its technology and application will reach the world's leading level;

By 2030, the theory, technology and application of artificial intelligence will reach the world's leading level and become the world's major artificial intelligence innovation center.

With the approach of the first deadline of 2020, how far has artificial intelligence developed in China?

Researchers have noticed an impressive leap in the quality of artificial intelligence research in China, and they also predict that China's ability to retain local talents will change.

However, observers warned that several factors may hinder China's plan, including a lack of contribution to developing tools to support theories in this field, and the reluctance of China enterprises to invest in the research needed to achieve fundamental breakthroughs.

Scientists say that China's pursuit of artificial intelligence is not only to compete with the national strength of the United States. Artificial intelligence technology is expected to make progress in medical care, transportation, communication and other fields, and countries that have made fundamental breakthroughs in this field may determine their future direction and gain the greatest benefits from it.

"There is no doubt that China regards artificial intelligence as one of the key technologies of this era, and hopes to rival the United States," said Jeffrey Ding, who studies the development of artificial intelligence in China at the Institute of Human Future of Oxford University.

After the release of 20 17 "A New Generation of Artificial Intelligence Development Plan", more policies and billions of dollars of R&D investment from ministries, provincial governments and private enterprises were promoted.

The quality of AI research in China has improved, but its core technology is backward.

The analysis of artificial intelligence papers collected by Microsoft Academic, an academic search engine, shows that China is steadily moving in the direction of great influence. According to the analysis conducted by Allen Institute of Artificial Intelligence, among the top 65,438+00% cited papers, the proportion of China authors has steadily increased. Its share reached the peak of 26.5% at 20 18, which was not far from 29% in the United States, while the share in the United States was declining. If this trend continues, China may surpass the United States in this indicator next year. Other analysis shows that the average citation rate of artificial intelligence papers in China has been increasing steadily, which is higher than the world average, but lower than that of American authors.

Please click to enter a picture description.

Zheng Nanning, director of the Institute of Artificial Intelligence and Robotics at Xi Jiaotong University, said that China also has world-leading companies in computer vision, speech recognition and natural language processing.

However, China still lags behind in building the core technical tools of artificial intelligence. For example, TensorFlow and Caffe, open source platforms developed by American scholars and enterprises, have been widely used in industry and academia all over the world. Zheng Nanning said that PaddlePaddle developed by Baidu in China is a major open source platform, which is mainly used for the rapid development of AI products.

China also lags behind in artificial intelligence hardware. Most of the world's leading artificial intelligence semiconductor chips are made by American companies, such as NVIDIA, Intel, Apple, Google and AMD. Zheng Nanning said: "We still lack the expertise to design computing chips that support advanced artificial intelligence systems."

Zheng Nanning predicted that it may take 5- 10 years for China to reach the innovation level of basic theories and algorithms in the United States and Britain, but this is an achievable goal.

Kristin Shi-Kupfer, a political scientist at the Mercator China Research Center in Berlin, said that contributing to these basic theories and technologies will be the key for China to realize its long-term goal of artificial intelligence. She said that if there is no real breakthrough in machine learning, China may face a growth ceiling in the field of artificial intelligence.

China has become more attractive to AI talents.

An equally important factor affecting China's progress is to retain talented researchers, and China seems more promising in this respect. According to the Report on the Development of Artificial Intelligence in China on 20 18, which was jointly written by academia and industry, by the end of 20 17, China had the second largest group of artificial intelligence scientists and engineers in the world, with about18,200 people, only about 29,000 people in the United States. But in terms of the number of top artificial intelligence researchers, China ranks only sixth. The so-called top is measured by their H index, that is, the most prolific and cited authors.

Please click to enter a picture description.

Many computer scientists receive higher education in the United States and then stay there to work for global technology companies.

However, there are signs that the situation is changing. Artificial intelligence research institutions in China are trying to attract some of them back to China with high salaries. He said, for example, in the Robotics Center in Zheng Nanning, some professors earn 2-3 times as much as other university professors.

Zheng Nanning said that the center also provides employees with a more comprehensive evaluation system than many universities in China. Compared with other standards, universities in China tend to reward high-risk models. He also implemented a recruitment system, bypassing the centralized procedures of universities, allowing scientists to quickly form a team of engineers, and is currently offering undergraduate courses in artificial intelligence.

China has a unique deployment environment.

Ding said that considering the growing professional skills of three core technology companies, Tencent, Baidu and Alibaba, China's plan to have a leading artificial intelligence company in the world by 2020 is also achievable. He said: "These companies have become global leaders in the field of artificial intelligence, although they have not yet reached the level of American companies such as Google and Microsoft."

According to CB Insights, there are at least 10 private AI start-ups in China with a valuation of over 1 100 million US dollars.

One advantage of China is its population size, which creates a huge potential labor force and unique opportunities for training artificial intelligence systems, including large patient data sets for training software to predict diseases. In February this year, researchers in China said that their NLP system can diagnose common childhood diseases from electronic health records with the same accuracy as experienced pediatricians. The data set includes the medical records of nearly 600,000 children in hospitals; In many other countries, it is difficult to get so much data.

China's artificial intelligence governance principles are taking shape.

If China wants to have global influence in the field of artificial intelligence, proper governance is equally important, because it will enable researchers and companies in China to build the necessary trust to win global users and establish cooperation with researchers from other countries.

Like many countries, China has begun to formulate ethical principles for the development and use of artificial intelligence. In June this year, the National New Generation Artificial Intelligence Governance Committee issued eight governance principles for the development of artificial intelligence, including harmony and friendship, fairness and justice, inclusiveness, respect for privacy, safety and controllability, responsibility, open cooperation and agile governance, similar to the measures issued by the Organization for Economic Cooperation and Development (OECD) in May this year.

To sum up, the quality of artificial intelligence research in China is getting higher and higher, the environment for applying and deploying AI is unique, and the ability to attract and retain talents is also improving, but China is still catching up with the United States in high-impact papers, talents and ethics.