Academician Bai Chunli: The next 5 to 10 years will be a critical period for the development of artificial intelligence

Yuanwang Think Tank foresees the future strategic frontier technology

Source: Sohu Technology

On May 17, the 2023 Sohu Technology Summit hosted by Sohu was grandly opened in Beijing. This year, the Sohu Technology Summit has entered its fifth year, and the specifications and content of the summit have once again ushered in a major upgrade. This summit gathered a number of academicians, scientists, scholars and guests from leading enterprises to discuss the development of cutting-edge science and technological change.

In the morning of the summit, Bai Chunli, chairman of the "Belt and Road" International Alliance of Scientific Organizations, former president of the Chinese Academy of Sciences and academician of the Chinese Academy of Sciences, delivered a speech entitled "Some Thoughts on the Development of Artificial Intelligence".

Bai Chunli believes that the next 5-10 years will be a critical period for the development of artificial intelligence. It is predicted that the scale of the core artificial intelligence industry in mainland China will exceed 1 trillion yuan in 2030, and the global artificial intelligence market will reach 16 trillion US dollars in 2030.

He said that artificial intelligence is a revolutionary and disruptive technology, which may trigger a new round of scientific and technological revolution, and then trigger an industrial revolution. From the perspective of R&D organization model, artificial intelligence requires a large-scale, high-level R&D platform to support a large amount of human and material resources. From the perspective of development drivers, the application orientation is very clear, and the pull of economic, social and other application needs has become the main driving force.

For the future development trend of artificial intelligence, Bai Chunli believes that basic theories and key technical fields are a large "no man's land", application systems are evolving in a more "intelligent" direction, and competition between system platforms will become the focus of wrestling. At present, major developed countries regard the development of artificial intelligence as a major strategy to enhance national competitiveness and maintain national security, and strive to take the lead in international scientific and technological competition.

Bai Chunli said that at this stage, mainland artificial intelligence needs to focus on solving three key problems: core technology and basic capabilities are not strong, industrial development ecology is not perfect, and high-end talents are seriously insufficient.

To this end, Bai Chunli put forward the following suggestions: First, give full play to the advantages of the new national system under the conditions of the socialist market economy, and concentrate the dominant cases to break through the key core technologies; The second is to strengthen the top-level design, strengthen overall coordination and system layout, and accelerate the construction of a benign interactive innovation ecology for government, industry, academia and research; The second is to innovate the talent training model, and accelerate the cultivation of a large-scale and first-class artificial intelligence talent team; Finally, it is necessary to strengthen the construction of laws, regulations, ethical norms and policy systems.

Bai Chunli is a former president of the Chinese Academy of Sciences

The following is the full text of Bai Chunli's speech:

When I met him, he had not yet founded Sohu, but he still maintains a strong interest in physics today and supports holding such a summit, which I think is particularly rare.

Just now there is a paragraph about artificial intelligence in the warm-up video, just now in the morning sun speech, mentioned the latest progress in artificial intelligence, today's thoughts on the development of artificial intelligence, including the development of disciplines and policy suggestions, to discuss with you.

In recent years, the rapid development of artificial intelligence has evolved from the field of scientific research to an enabling body that can profoundly change and affect various industries, and as an infrastructure to promote economic and social development, showing broad application prospects and attracting great attention from all aspects.

First, on the leadership and strategic dominance of artificial intelligence. With its technical characteristics, depth and breadth of influence, and radiation driving effect, artificial intelligence is expected to become a strategic key technology with a leading and dominant position in the new round of scientific and technological revolution.

From the historical perspective, the distinctive characteristics of the historical scientific and technological revolution and the industrial revolution are all around human development and liberation. The first industrial revolution was marked by the steam engine, which liberated people from heavy physical labor, and the second industrial revolution, marked by the internal combustion engine of electric technology, greatly surpassed human physical ability and met the diversified needs of human work and life. The third industrial revolution marked by the computer technology Internet has realized the efficient replacement of simple mental work, and greatly expanded the space for human development, liberating human beings at a higher level and higher scope. The third industrial revolution has completed the expansion and liberation of human physical and mental strength, and artificial intelligence will realize the imitation, replacement and expansion of human complex intellectual and intelligent activities, and liberate human beings at a higher level and in a larger scope.

From the perspective of development, artificial intelligence is a revolutionary disruptive technology, which may trigger a new round of scientific and technological revolution, and then research and development industrial revolution. Artificial intelligence is an important force in the new round of scientific and technological revolution and industrial revolution, which can accelerate the new generation of artificial intelligence, and is a strategic issue for the opportunity of a new round of scientific and technological revolution and industrial transformation. We also know that if any current type of technology can really be compared with the leading technology in the previous industrial revolutions, it is undoubtedly artificial intelligence.

Therefore, artificial intelligence integrates all scientific and technological innovation achievements since mechanization, automation and informatization, which will lead economic and social development into a new era full of infinite possibilities, and will profoundly affect and change the pattern of global economy, industry and innovation.

At present, developed countries in the world regard the development of artificial intelligence as a major strategy to enhance national competitiveness and maintain national security, and strive to seize the initiative in international scientific and technological competition. For example, in 2019, the United States issued the Executive Order on Ensuring American Artificial Intelligence Leadership, launching the National Institute of Artificial Intelligence program. By the end of 2021, a total of 18 national AI research institutes had been funded, and the 21st National AI Initiative Act last year clearly listed AI as a key technology field critical to maintaining national security.

Similarly, the UK has also established the Turing Institute, committed to making the UK a global leader in the field of big data analysis and applications, and in April 2019, 4 artificial intelligence interdisciplinary institutes were established, involving health, environment, energy, transportation and other fields.

Second, the main characteristics and development trends of artificial intelligence.

In 1956, Minsky and other scientists proposed the concept of artificial intelligence at the Dartmouth Conference, marking the birth of the discipline of artificial intelligence, after more than 60 years of development, the development of artificial intelligence has been ups and downs, and has shown a vigorous and rapid development trend in recent years. Our Sohu Technology Summit has been dedicated to the topic of artificial intelligence development. ChatGPT represents a new round of artificial intelligence boom, from the perspective of the internal attributes of the discipline, artificial intelligence multidisciplinary cross-integration characteristics are particularly obvious, and its technology is very comprehensive, research and development is also very difficult, artificial intelligence set mechanization, intelligence, information technology into the technology, high-level system research and development first rely on computers, integrated circuits, network communications, advanced computing, mechanized structures, sensors, key materials and other fields of direct support. The subversive progress of intelligence often requires theoretical breakthroughs in basic disciplines, such as algorithms. Focusing on simulating and surpassing human intelligence, the integration of artificial intelligence with brain science and cognitive science research is becoming closer and closer.

From the perspective of R&D organization model, artificial intelligence requires large-scale, high-level R&D platform support, and also requires a large amount of human and material resources.

The intrinsic nature of the discipline of artificial intelligence determines that only large teams and large platforms with deep cross-integration of multiple disciplines can achieve great breakthroughs. This trend is particularly obvious in this round of development wave, if in the development before, very talented scientists or teams can play a decisive role, but this round of major breakthroughs, are large institutions and large enterprises, such as Google and Microsoft, IBM's artificial intelligence team are thousands of people, large platform, large team behind the support of large investment, this is the organizational model of research and development, just talked about the intrinsic attributes of the discipline, from the perspective of the research and development organization model.

From the perspective of development drivers, I think that the application orientation of artificial intelligence is very clear, and the demand for economic, social and other applications has become the main driving force.

In this round of artificial intelligence development, the first two rounds are largely driven by technology, in this round of development, although technological progress also plays an important role, but the main driving force is from the continuous expansion of application scenarios, and the deep integration of economic and social applications, thus spawning a large number of new industries, new formats, such as the ChatGPT developed by OpenAI in the United States, its development has indeed attracted widespread attention around the world.

The next 5-10 years will be a critical period for the development of artificial intelligence, and according to statistics in 2021, the scale of the core industry of artificial intelligence in mainland China will be larger, more than double the same period in 2019. According to forecasts, the scale of the core artificial intelligence industry in mainland China will exceed 1 trillion yuan in 2030, and the global artificial intelligence market will reach 16 trillion US dollars in 30 years, and there is still great potential and space for development.

First, in the field of basic theory and technology, there is a large area of no man's land, about generative artificial intelligence, brain-like intelligence, perception and chips, perception devices, all need continuous breakthroughs, in order to achieve the ultimate goal of general artificial intelligence, this aspect also needs more forward-looking and even unknown theoretical breakthroughs.

Second, the evolution of the application system to be more intelligent. Artificial intelligence will gradually reduce the dependence on artificially designed network models, manually labeled big data, time-consuming and labor-intensive training, through sample input, independent learning and evolution of external environment perception, and the ability to adapt to different tasks in different environments, from the original artificial + intelligence, to autonomous intelligence, human-computer collaboration, brain-computer interface. From machine intelligence to human-machine hybrid intelligence, this is an inevitable trend in the development of artificial intelligence, and it is also a revolutionary leap of challenges.

Third, competition between system platforms has become the focus. The competition in the field of artificial intelligence, but the competition of the big system, the big army, the competition for massive data resources and users, the development focus gradually shifted to the open source ecosystem, thereby forming new technical barriers for latecomers, such as Google, IBM, Microsoft, Facebook and other technology giants actively promote the construction of artificial intelligence ecology, seize the commanding heights of open source platform competition. I've illustrated this trend.

Finally, I would like to talk about my thoughts and suggestions on the development of artificial intelligence in the mainland.

The central government attaches great importance to the development of artificial intelligence, and to strengthen the development of artificial intelligence, it is required to deeply understand and accelerate the great significance of the development of a new generation of artificial intelligence, strengthen guidance, clarify tasks, consolidate the foundation, promote the integration of economic and social development, and promote the healthy development of the new generation of artificial intelligence in the mainland. The State Council specially formulated and issued the "New Generation of Artificial Intelligence Development Plan", which carried out top-level design and overall layout of artificial intelligence innovation chain, industrial chain, application chain and other dimensions, which also provided strong policy guidance for the accelerated development of artificial intelligence.

Overall, the mainland has invested relatively heavily in artificial intelligence, the number of high-level papers and core patents is increasing rapidly, and some artificial intelligence leaders are developing rapidly. In addition, we have a huge market scale and market potential, the most complete intelligent manufacturing and industrial chain, the largest massive data resources, through the "intelligence +" action, the promotion of artificial intelligence technology penetrates into all walks of life, forming a large-scale intelligent application market, I think this is a unique advantage of the mainland to develop artificial intelligence. Because our market is very big.

In the future, the competition in the field of artificial intelligence will become more and more fierce, in order to ensure the rapid development of the new generation of artificial intelligence in the mainland, at this stage, efforts are made to solve three key problems.

1. The core technology and basic ability are not strong. The mainland has entered the original innovation and technology accumulation in the field of artificial intelligence, and the basic theoretical framework and core key components such as chips rely on imports more, which may lead to the mainland being "hollowed out" again in the intelligent stage. This is where the biggest gap between us and the United States lies.

2. The industrial development ecology is not perfect. The development of independent technology system and industrial ecology is weak, foreign countries have built up the barriers of the ecosystem by using technical barriers and first-mover advantages, a large number of market applications and development are uneven, there are more low-end, homogeneous competition is obvious, and the market competition environment still needs to be improved. Artificial intelligence involves ethical issues and social issues research is also relatively lagging behind, and legislation and governance need to be strengthened.

3. There is a serious shortage of high-end talents. Tsinghua University and other units released the 2022 world's most influential scholars, AI2000 list, this list in the United States is the most, 1164 people selected, China has 332 people selected, the third is the United Kingdom 115 people selected, it can be seen that the mainland AI universities are still insufficient talents, many universities have opened artificial intelligence majors, not long ago I attended the State Council Ministry of Education on artificial intelligence discipline attention, lack of high-quality teachers, curriculum and market demand do not match, The scale and quality of talent training are difficult to meet the development needs.

Based on the above analysis of the disciplinary characteristics, development trends and existing problems of artificial intelligence, there are the following suggestions:

Suggestion 1: Give full play to the advantages of the new national system under the conditions of the socialist market economy, and concentrate advantageous cases to break through key core technologies. Give full play to the demand-side traction role of the huge market, fully stimulate the vitality and potential of enterprises as the main body of technological innovation, build an independent platform ecology, and form a good development situation in which scientific and technological innovation and industrial application promote each other. In particular, leading enterprises in the industry have increased investment and joined forces to tackle key problems.

Give full play to the institutionalized and multidisciplinary advantages of the national strategic scientific and technological forces, grasp the layout on the short board, and ensure that the key core technologies are firmly in their hands.

Suggestion 2: Strengthen top-level design, strengthen overall coordination and system layout, and accelerate the construction of a benign interactive innovation ecology for government, industry, academia and research.

We know that "two bombs and one satellite", this is a major achievement of the national system, compared with two bombs and one satellite, and then look at the strategic tasks of space countries, the field of artificial intelligence, involving multiple disciplines and their cross-integration, involving a wider range of more complex system engineering. It not only involves multidisciplinary and cross-integration, software and hardware systems such as chips and operating systems, complex open source platforms, massive data resources and standard system construction, interaction between technology and economic and social applications, cultivation of industrial competitiveness and military-civilian collaboration under the conditions of market economy, international competition and cooperation, and a large number of talent teams in different disciplines.

Therefore, in this regard, it is especially necessary to strengthen top-level design and comprehensive policies, and it is necessary to coordinate and cooperate with various policies such as projects, bases, talents, and education in order to play a better role. For example, it is necessary to lay out national artificial intelligence research centers and laboratories, build independent and controllable national open source platforms, etc., including open source of data, management of policies, and reduction of data islands.

Suggestion 3: To innovate the talent training model, we must accelerate the cultivation of a large-scale and first-class artificial intelligence talent team.

It is necessary to combine the characteristics of artificial intelligence disciplines and the current problems, build a high-level talent training system based on independence, and make overall considerations, systematic design education and training, talent plans, talent evaluation and other policies.

In terms of high-level personnel training, speed up reform and improve the talent evaluation and incentive system. Establish a benign competition mechanism and a reasonable flow mechanism for talents. Create a creative environment conducive to breakthrough and disruptive innovation, and strengthen continuous and stable support for basic research and key core technology research.

In terms of young talent training, integrate excellent teachers with relevant disciplinary backgrounds, combine major projects and important application practices, and rely on high-level platforms to cultivate interdisciplinary compound talents.

Recommendation 4: Strengthen the construction of laws, regulations, ethical norms and policy systems.

The development of artificial intelligence has caused widespread concern in the scientific community, Elon Musk thousands of experts, calling for a suspension of AI research for 6 months, how everyone responds is one thing. And we know that calling on countries to immediately implement the Recommendation on Ethical Issues of Artificial Intelligence, the Cyberspace Administration of China issued the "Draft for Comments" of the Measures for the Management of Generative Artificial Intelligence Services, so in view of the security risks that artificial intelligence may bring, including technical forgery, some people are worried that artificial intelligence falsifies information, and it is difficult for everyone to identify.

There is also an invasion of privacy. I see that South Korea does not allow the use of ChatGPT, that is, when the conversation occurs, I am afraid that the information will be exposed. There are also social ethics, replacing artificial labor to change the employment structure, and how to formulate international rules for artificial intelligence, etc., which need to be carefully studied. To accelerate the construction of a more inclusive and prudent domestic regulatory system, on the one hand, we must supervise, but also encourage and support the rapid development of artificial intelligence, and at the same time, we in China also need to strive to have a greater say in the formulation of relevant international rules. We need to work harder in this regard.

I'll stop here today. Thank you!