撰写August 29th, 2019WAIC World Artificial Intelligence Conference — — International conference on frontier information of artificial intelligence was held in Shanghai. Around the theme of "bottleneck and breakthrough of artificial intelligence development", research institutions, enterprises and governments from home and abroad have conducted exchanges and discussions. The discussion holds that AI has made great progress in the past ten years, but the technology is still in a relatively early stage; China’s artificial intelligence pays more attention to the application level. Regarding the empowerment of artificial intelligence for the industry, many participants believe that artificial intelligence should not only pay attention to the upgrading of hard technology, but also constantly improve the construction of "soft science" such as law, culture and ethics behind it.
The 2019 World Artificial Intelligence Conference was opened, and the staff (right) was demonstrating a robot with remote control based on 5G network. Xinhua News Agency reporter Fang Yushe
The development of artificial intelligence is in its infancy, and the bottleneck needs to be broken.
With the exploration of driverless technology in the automobile industry, industrial and commercial development is increasingly interested in automation, and people are increasingly hopeful that artificial intelligence will replace manual operation in many aspects in the future.
"At present, artificial intelligence is only an initial stage of learning and training. In the future, we need to be able to make intelligent decisions. In fact, what we want to do now is to lead and control the system by people, and technology is also very important, "said Professor Carles Sierra, a member of the International Artificial Intelligence Council and a member of the Spanish National Research Council." Even if a complete automation is realized, we also hope that it must be led by people, so that some risks can be avoided. This is the main bottleneck of our AI development at present. In other words, in the future development process, we must ensure that it is led by people to carry out risk management and control. We don’t have a good solution yet, that is to say, in the process of autonomy of our large automation system, we need people to cooperate. "
AI efficiency and scalability are bottlenecks.
Michael Wooldridge, director of the Department of Computer Science at Oxford University and academician of the European Academy of Sciences, believes that "its efficiency and scalability now include some requirements such as computing resources, which are bottlenecks. It is also urgent for us to find a solution, especially in terms of efficiency improvement. We hope to really conduct in-depth research on this kind of machine learning. Especially in the aspect of concept cognition, it can recognize words and sounds, and it also has the ability of reasoning. In the past ten years, artificial intelligence has developed many technologies, and we must find a correct way to carry out such applications. "
Rao Kotagiri, an academician of the Australian Academy of Technology and Engineering and a professor at the University of Melbourne, believes that "the current AI system is actually relatively low-level and primitive. We need to make it more advanced and advanced, and we need more relevant background information and more logic. I hope that AI can be expanded, and we hope to provide some AI system solutions comprehensively. To ensure its security, so that AI can ensure safety when doing better than humans. "
Technical ethics needs attention.
The US Federal Trade Commission previously investigated Youtube’s alleged violation of children’s privacy laws and issued a $200 million fine to the parent company Google. Zao, an AI face-changing software, was popular on the Internet for a while, and correspondingly, users’ privacy security and data security risks caused thinking. These events reflect that the "soft package" of law, culture and ethics that is suitable for technological development is not perfect. How to effectively supervise artificial intelligence technology, ensure technical safety, and at the same time effectively promote the market-oriented development of artificial intelligence is a major challenge facing the development of artificial intelligence.
No matter which industry is empowered, the issue of public rights and interests is an urgent problem to be solved in the development of artificial intelligence. Sun Jianping, dean of the Urban Risk Management Research Institute of Tongji University, stressed, "At present, we are still in the exploration stage of artificial intelligence technology, such as face recognition technology applied in the security field. If it is not properly kept and clearly used, the abuse of data such as face images or videos will directly infringe on the public’s rights and interests."
"Relatively speaking, the United States pays close attention to the risks related to artificial intelligence in the development of artificial intelligence, while China pays more attention to the application of technology. The United States has fully discussed the risks brought by artificial intelligence to society, and established a safety and ethical risk prevention and control mechanism with various comprehensive deployments, while China’s emphasis on risk safety needs to be further improved. " Professor Yin Feng from Shanghai University said.
Technology research and development, perfect legislation and supervision, and jointly break through the bottleneck of development.
Technological development is an important driving force to break through the bottleneck. In addition to enhancing the hard power, Professor Wang Yansong, vice president of Shanghai Engineering Universities of Applied Sciences, believes that "while enhancing the hard power of AI, we should also pay attention to the research of artificial intelligence-related soft disciplines, such as the technology roadmap, business model and applied ethics for the development of artificial intelligence industry. There is also the deep integration of artificial intelligence and traditional industries and the resulting revolutionary problems in management theory. "
Perfecting legislation and strengthening supervision are also important means to escort the development of AI. "We need a corresponding legislative system and a punishment mechanism for misconduct. At the same time, we must also conduct corresponding research from the theory of economics. " Carles Sierra said.
The Three-year Action Plan for Promoting the Healthy Development of a New Generation of Artificial Intelligence (2018-2020) promulgated by the Ministry of Industry and Information Technology in 2017 puts forward requirements for the development and governance of artificial intelligence, but the implementation of rules and requirements is still in progress. Fu Ying, former deputy minister of the Ministry of Foreign Affairs and director of the Center for International Strategic and Security Studies in Tsinghua University, mentioned in his speech on "Governance and International Cooperation of Artificial Intelligence" that "the important task now is how to make these rules and requirements fall into place and form a benign environment conducive to the healthy growth of artificial intelligence in China."
China artificial intelligence pays more attention to the application level.
"Compared with the United States, the advantages of artificial intelligence development in China are more reflected in the application level," Yin Feng said. "At the same time, there are many problems in the application and promotion level."
Dr. Liu Bing, Chairman of Shanghai Zhongyi Energy Technology Co., Ltd. believes that "China’s advantage lies in the shaping of traditional culture and application scenarios, while the West lies in the integration of basic research and global resources. If we can explore a methodology, we can combine China’s wage socialization needs with the governance model of big data, and be both wise and brave. How can we combine data from top to bottom and form a good prescription for vulnerable groups through this governance of data and thousands of solutions?"
The road to true artificial intelligence is still far away
The application of artificial intelligence has expanded to many fields. Intelligent education products have helped teachers improve the efficiency of classroom teaching in thousands of schools. Artificial intelligence general practitioners can help doctors diagnose nearly a thousand diseases, significantly improve the level of primary medical care and promote medical equity. More and more urban brains are accumulating data and obtaining more efficient and intelligent urban management practices.
In particular, artificial intelligence is widely used in the practice of urban management to effectively cope with the pain points and problems of urban operation. With the gradual development of artificial intelligence, many products and technologies are named as artificial intelligence, but the practical application still stays in the case link.
"In fact, artificial intelligence is far from real intelligence, and the example of AlphaGov can only realize real artificial intelligence on a specific occasion and in a specific scene. So the overall distance of artificial intelligence is still far away. " Professor Wang Yansong believes that "as a research institution such as a university or research institute, it is more necessary to do deeper research from the basic theory."
Artificial intelligence application dividends should take root and be inclusive.
How to make good use of artificial intelligence to run a satisfactory education for the people? Zhou Hanmin, member of the Standing Committee of Chinese People’s Political Consultative Conference, vice chairman of the Central Committee of Democratic National Construction Association and vice chairman of Shanghai CPPCC, believes that artificial intelligence and education in China are related to body and use. The former is for use and the latter is for body, which cannot be reversed or ignored.
In view of the fact that there are three characteristics of education in China: weak countryside, crowded cities and weak teachers, Zhou Hanmin put forward three suggestions on the combination of artificial intelligence and education. One is to establish a national online compulsory education platform, so that the whole education system can be moved online by high-quality teachers; Second, the Internet project and education project should be carefully combined to promote the spread of the Internet in all schools; Third, we should strengthen international cooperation in education. Zhou Hanmin said, "The effect of the combination of the two is not how high-tech can fly higher and go further, but how high-tech means can take root and benefit the people."
Similarly, AI technology has made some achievements in the field of empowerment medicine, which not only provides support for medical research, but also brings a lot of help to the medical service society.
In 2017, Nature magazine published an article on the cover: using deep learning algorithm to diagnose skin cancer, with an accuracy of 91%, which can be compared with doctors. "It may take 30 years for dermatologists to be able to work, and artificial intelligence machines only need 24 hours of training to have the same diagnosis." Rao Kotagiri, a professor at the University of Melbourne and an academician of the Australian Academy of Technology, Science and Engineering, believes that "outsiders question that AI will make mistakes, but this technology has helped those countries that do not have dermatologists."
"Artificial intelligence can’t completely replace human beings, but it can replace many routine and repetitive transactions, leaving the transactions that really need experts to handle to the end. China’s medical staff are very nervous now, and AI can do some real auxiliary work, which is what we hope. " Li Junmin, director of hematology department of Ruijin Hospital affiliated to Shanghai Jiaotong University, believes that "artificial intelligence is more for social service. Especially in some areas, there is no such technology, and artificial intelligence really benefits the grassroots. "
How to fully apply artificial intelligence in various fields? Fang Ming, assistant to the president of Iflytek Co., Ltd., believes that "artificial intelligence must meet three conditions to realize the value dividend. First, it must have real and visible application cases. Second, it must be able to be promoted on a large scale. Finally, it must be able to prove the real application effect with statistical data."
This conference is directed by the Organizing Committee of the World Artificial Intelligence Congress, and co-sponsored by China Academy of Social Sciences-Shanghai Research Institute of Shanghai Municipal People’s Government, Shanghai Committee of China Democratic National Construction Association, Shanghai University, Shanghai University of Engineering Science and Shanghai Investment Promotion Center of the United Nations Industrial Development Organization.