The advent of artificial intelligence models such as ChatGPT has triggered a wave of technological change. This weekend, the "Pujiang Innovation Forum - 2023 Science and Technology Innovation Think Tank International Seminar" with the theme of "AI Reshapes the Paradigm of Science and Technology Innovation" was held in Shanghai, and experts from first-class science and technology innovation think tanks and industry-university-research institutions at home and abroad delivered speeches.
Zhang Xu, President of the China Academy of Science and Technology Development Strategy, said: The next 10 years will be an important period for disruptive technologies such as artificial intelligence, quantum computing, and controllable nuclear fusion to empower each other and achieve group breakthroughs, and it will also be a key decade to determine the pattern of competition among major countries.
Three disruptive technologies empower each other
"There are two tech news stories worth watching lately." Dean Zhang Xu said in his speech that first, in May this year, Microsoft signed a power purchase agreement with Helion Energy, and will purchase 50 megawatts of electricity from the latter in 2028. The second news is related: OpenAI founder Sam Altman invested $375 million in Helion Energy, and the two companies will work together to accelerate experiments on controlled nuclear fusion with AI models.
Dean Zhang Xu delivered a speech at the seminar.
The AI models such as ChatGPT and GPT-4 developed by OpenAI have been known to the public. What is the technology for controlled nuclear fusion developed by Helion Energy? According to reports, controllable nuclear fusion is commonly known as "artificial sun", that is, simulating the nuclear fusion process of the sun, using the isotope of hydrogen - the fusion reaction of deuterium and tritium to release energy. It is regarded as a "ultimate solution" to solve human energy and carbon emission problems, because deuterium has up to 40 trillion tons of reserves in the global ocean, and if all of it is used in fusion reactions, it will release enough energy for unlimited human use, and its reaction product is helium without radioactive pollution, and it will not produce greenhouse gases such as carbon dioxide.
Although the application prospects are promising, the technical difficulty of controllable nuclear fusion is extremely high, and a major problem is the "tokamak magnetic control problem". A controlled fusion device called a tokamak that confines the high-temperature plasma so that the fusion reaction can be sustained and controlled. With the increase of nuclear fusion reactors, tokamak devices become more and more complex, and the operational settings of their internal configuration are becoming more and more demanding, and this control process is nonlinear, real-time changing, and multivariate, which traditional calculation methods cannot afford.
In February last year, artificial intelligence company DeepMind, in cooperation with the Plasma Center of the Ecole Polytechnique Fédérale de Lausanne, published a paper in the journal Nature entitled "Magnetron of Tokamak Plasma through Deep Reinforcement Learning", showing that artificial intelligence technologies such as deep learning and reinforcement learning can help achieve high-precision control of tokamak devices, making human dreams of "artificial sun" come true as soon as possible.
Fully superconducting tokamak nuclear fusion experimental device (Xinhua News Agency)
In the cutting-edge field of quantum computing, artificial intelligence also has important value. It can optimize the design of hardware and software in quantum computing applications, and improve the efficiency and accuracy of quantum technology applications. "There is a mutually reinforcing and interdependent relationship between artificial intelligence and quantum computing. In the future, with the development of science and technology in these two fields, this relationship will become closer and more complex. Zhang Xu said.
Fifth, it is recommended to seize the historical opportunity
Looking forward to the next 10 years, disruptive technologies such as artificial intelligence, quantum computing, and controllable nuclear fusion will achieve and iteratively break through each other. Can the mainland seize this period of historic opportunity? As an expert in science and technology strategy, Zhang Xu put forward 5 suggestions.
The first is to coordinate the layout of energy bases and computing power to form the basic competitiveness of artificial intelligence development. Compared with other countries, a unique advantage of the mainland is the cool climate in the west and abundant and clean power resources. It is necessary to make full use of this advantage, promote projects such as "East Data and West Computing", build a number of computing power factories in the west, form a low-cost computing power network combining supercomputing, intelligent computing and even future quantum computing, and provide competitive basic conditions for the development of artificial intelligence and other technologies.
The second is to simultaneously support diversified technical routes and form the ability to cope with rapid technological changes as a whole. The development of disruptive technologies is "emerging", and it is difficult to predict which technology route will win and which company will win. Before the route is clear, multiple scientific and technological forces in different directions should be laid out, not only adhering to original innovation, but also maintaining the ability to quickly follow up any technical line.
The third is to actively design extremely challenging scenarios and guide more long-term and original scientific and technological research. Major missions such as lunar bases, Mars exploration, and deep-sea exploration will bring exciting scenarios to the development of strong artificial intelligence, and long-term goals and important milestones should be set to encourage scientists to carry out long-term and original research.
The fourth is to strengthen enterprise-centered collaboration and improve the control of disruptive technology ecosystems. Enterprises are sensitive to the market, have strong ability to allocate resources, and can best seize the opportunities of the scientific and technological revolution. In the United States, the dominant forces in artificial intelligence, quantum computing, and controlled nuclear fusion are all enterprises. A number of world-class science and technology leaders have emerged on the mainland, and it is necessary to guide enterprises to strengthen basic research, form synergies with high-level universities and scientific research institutions, and strengthen the alliance between enterprises to build a disruptive technological innovation ecology with leading enterprises as the core.
Fifth, do everything possible to promote international exchanges and cooperation, introduce more high-level talents, and deeply integrate into the international innovation ecology, so that the mainland will become a global hub for disruptive technological change.
"Solutions to the Changing Situation: Thought Observations of Global Science and Technology Think Tanks" released.
At the seminar, the "Solution to the Changing Situation: Ideological Observations of Global Science and Technology Think Tanks", led by the Shanghai Institute of Science and Technology, jointly completed by Fudan University, Tongji University, Chinese Academy of Sciences, Shanghai Science and Technology Management Cadre College, Moscow State Lemmonosov University and other institutions, was also released. The first part focuses on the research results of 12 representative science and technology think tanks in the world in the past five years in promoting the development of science and technology, promoting the application of science and technology, strengthening science and technology governance, and helping scientific and technological competition between countries, and the second part focuses on analyzing the representative topics of global science and technology think tank research, aiming to provide reference for the development and scientific decision-making of mainland science and technology think tanks through the analysis of science and technology think tank issues and programs.
Editor-in-chief: Huang Haihua
Source: Author: Yu Taoran