Shulou(Shulou.com)06/02 Report--

According to a recent analysis by Nature, China's artificial intelligence research has made rapid progress in quality, but it still needs to catch up with the United States in terms of high-impact papers, talent and ethics.

Selected from Nature, author: Sarah O'Mear, compiled by Machine Heart.

China's huge population has contributed to the rapid progress of face detection technology.

China is not only the most populous country in the world, but also seems to be on the verge of becoming the country with the largest economy. In the field of artificial intelligence, China also seems to want to lead the world.

In 2017, China released the "New Generation artificial Intelligence Development Plan", which set the AI development goal with 2030 as the deadline. In order to achieve this goal, there are many milestones to be achieved by 2020, including making significant contributions in the field of basic research, becoming a destination for attracting the world's emerging talent, and reaching the world's leading level in the artificial intelligence industry. After the plan came out of Taiwan, central departments, local governments and private enterprises responded positively, attracting billions of dollars in financing for artificial intelligence research and development.

As this first deadline is approaching, the researchers have noticed that the quality of artificial intelligence research in China has made amazing progress. They expect China to be able to retain local talent. Part of the reason is that the government has implemented some successful talent introduction programs. Part of the reason is that deteriorating diplomatic and trade relations have made the United States, China's main competitor, a relatively unattractive destination in all aspects, including artificial intelligence.

"if the United States is no longer at the forefront of opening up, it is risking sending artificial intelligence talent to major competitors, including China. 喜悦 Dantong Ma, deputy director of the Marco Polo think tank of the Paulson Foundation in Chicago, USA.

The Marco Polo think tank, an internal think tank of the Paulson Foundation, aims to interpret the Chinese economy with keen market analysis and innovative research.

But at the same time, observers warn that some factors may hinder China's national strategy, including the lack of theoretical contributions that can guide the development of the field and investment in basic scientific research from Chinese companies.

China's pursuit of artificial intelligence is more like a formal manifestation of competition with the United States, some scientists say. Artificial intelligence can promote the development of medical, transportation and communication fields, and countries that have made fundamental breakthroughs in this field will lead the future development direction and gain the greatest benefits.

There is no doubt that China regards artificial intelligence as a core technology and wants to reach the level of the United States. "said Jeffrey Ding of the Institute for the Future Humanities (the Future of Humanity Institute) at Oxford University.

Academic influence

A Microsoft academic study of highly quoted artificial intelligence papers shows that China's influence is very strong. Analysts from the Allen Institute of artificial Intelligence found that Chinese authors increased their share of the top 10% of highly cited papers by 10%, peaking at 26.5% in 2018, close to 29% of those in the United States, and the share in the United States is falling. If this trend continues, China will overtake the United States in the second year (2019). Other analyses show that the average citation rate of artificial intelligence papers by Chinese researchers has increased significantly, which has exceeded the world average, but is still lower than that of American researchers.

The citation level of Chinese artificial intelligence papers is increasing year by year, which is close to that of the United States.

The influence of Chinese universities in areas such as artificial intelligence has also continued to grow in the computer science ranking of top schools (CSRankings). The ranking is based entirely on research indicators and measures the number of papers published by the vast majority of faculty in the field of computer science.

In addition to China's growing academic influence, China's artificial intelligence industry is also booming. Professor Zheng Nanning, academician of the Chinese Academy of Engineering and director of the artificial Intelligence and Robotics Center of Xi'an Jiaotong University, said that China has a number of the world's top artificial intelligence companies, such as Shangtang, Yunzhisheng, iFLYTEK and Xiangshi. They have made great achievements in computer vision, speech recognition and natural language processing.

Of course, China still lags behind in the tools to build the core technology of artificial intelligence. For example, open source platforms such as TensorFlow and Caffe are designed by American academic institutions and companies. These open source platforms can help computers work more like the human brain, so they are widely used in industrial and academic fields around the world. Of course, Baidu Flying Propeller platform is the most frequently used in the rapid development of artificial intelligence products, said Professor Zheng.

Professor Zheng also mentioned that China also lags behind in artificial intelligence hardware. Most of the world's leading artificial intelligence microprocessor chips are developed by American companies such as Nvidia, Intel, Apple, Google and AMD. "We also lack expertise in designing computer chips for advanced artificial intelligence systems. "

Professor Zheng predicts that it will take China five to 10 years to reach the level of basic theory and algorithm research in Britain and the United States, but China will do so sooner or later.

Being able to contribute basic theory and technology is the key to China's long-term artificial intelligence strategy, said Kristin Shi-Kupfer, a policy scientist from a Berlin think-tank. She said that if there is no breakthrough in machine learning, China may face the ceiling in the field of artificial intelligence.

Talent situation

How to retain talents is another difficult problem for China's development of AI. According to the 2018 China AI Development report jointly written by academia and industry, by the end of 2017, the number of AI researchers and engineers in China had reached 18200, ranking second in the world. However, China only ranks sixth in the ranking of top AI researchers (highly productive and highly cited researchers).

In a report on the comparison of AI strength among China, the United States and Europe released by Center for Data Innovation, a cross-research think tank on data, technology and public policy in the United States, the situation of AI talents in China is also worrying. As of 2017, China's top AI talent (the top 10 per cent of the h index) was less than that of the US, according to the report.

Ma says many computer scientists are usually educated in the United States and will stay there after graduation to join some of the world's top technology companies.

But there are signs that the situation is improving. AI institutions in China are trying to attract these researchers back home with high salaries. For example, at Professor Zheng Nanning's robotics center, some professors are paid two to three times as much as others. In addition, China's education system has also stepped up the training of AI talents. last year, 35 universities were approved as "artificial intelligence" undergraduate majors, and famous universities such as Tsinghua University have set up a number of AI research and talent training centers to solve the shortage of AI talents in China.

Professor Zheng added that his robot center also provides a more comprehensive evaluation system to improve talent attraction. In contrast, many Chinese universities tend to use the number of papers published as a reward standard. In addition, he has created a recruitment system that can bypass the process of university centralization and help scientists build a team of engineers more quickly, which is now conducting undergraduate courses in AI.

China's Enterprise and population advantages in developing AI

Ding said that given the increasing expertise and industry influence of the three core technology companies, Tencent, Baidu and Alibaba, it is believed that China's plan to have the world's leading AI company by 2020 can be achieved. "although they have not yet reached the level of US technology giants such as Google and Microsoft, these three companies have become global leaders in AI," he said. "

There are at least 10 AI venture capital companies valued at more than $1 billion in China, according to CB Insights, a venture capital research firm in New York.

In addition, although it is inferior to the United States in terms of venture capital and private equity financing, China has the highest percentage of successful incorporation of AI into the company's business processes. In 2018, China led the world in this respect (32 per cent), higher than the United States (22 per cent) and the European Union (about 18 per cent). On the other hand, 53% of Chinese companies are already piloting artificial intelligence applications, which is also well ahead of the second place in the United States (29%).

Ma said that one of China's major advantages is its huge population, which provides sufficient personnel samples and unique opportunities for training AI systems, such as the large patient data sets needed to train disease prediction software, which can bring great advantages to AI research and development.

Compared with Western countries, Chinese Internet companies have the opportunity to collect more diverse and in-depth data than American Internet companies. This is thanks to integrated super applications built by Chinese technology companies. For example, Wechat now allows users to "call taxis, book meals, book hotels, pay for mobile phone calls, and buy flights to the United States." By contrast, Internet services in the West are relatively fragmented: even Amazon users cannot book hotels.

Of course, China also has a disadvantage in terms of data. American technology giants have a wider global reach, so the scale of global users is unparalleled. For example, Facebook has more than 2 billion users, while Wechat has only 1.1 billion users. If Chinese companies achieve greater international success, as Douyin, a short video app, does, America's advantage will diminish.

AI governance and the formulation of AI principles are imperative

Ma said that if China wants to have an international impact in the AI field, it is also important to implement appropriate governance, so that Chinese AI researchers and companies can build the necessary trust to win users around the world and cooperate with researchers in other countries. Like other countries, China has begun to formulate ethical norms for the development and application of AI. Therefore, Chinese AI companies need to commit themselves to good governance in order to obtain global data.

In June this year, China's National New Generation artificial Intelligence Governance Professional Committee issued the "New Generation artificial Intelligence Governance principles-developing responsible artificial Intelligence" that AI developers should abide by, put forward the framework and guidelines for artificial intelligence governance, and clearly put forward eight principles such as harmony and friendship, fairness and justice, tolerance and sharing, respect for privacy, security and controllability, shared responsibility, open cooperation, and agile governance. These eight principles coincide with the principles of artificial intelligence, which were voted by the Organization for Economic Cooperation and Development in May and adopted by governments around the world.

However, it should be noted that different countries face different ethical problems. Therefore, each country will formulate relevant artificial intelligence standards according to its own national conditions and actual situation.

In addition, another key challenge facing all countries and regions is the transparency of algorithms in decision-making. There is no uniform standard in this regard, so China and other countries are still exploring how to move the process forward. For example, the EU's outgoing General data Protection Regulation (General Data Protection Regulation) gives users the right to ask how algorithms make decisions when it comes to themselves.

Reference link:

Https://www.nature.com/articles/d41586-019-02360-7

Https://www.datainnovation.org/2019 / 08/who-is-winning-the-ai-race-china-the-eu-or-the-united-states Who Is Winning the AI Race: China, the EU or the United States? Https://www.datainnovation.org/2019 / 08/who-is-winning-the-ai-race-china-the-eu-or-the-united-states

Https://zhuanlan.zhihu.com/p/79290066

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.