Is artificial intelligence giving rise to a new "technological revolution"?

ChatGPT, which was born out of nowhere, has gone from being favored and sought after at the beginning to being questioned and cautious now, but it has not reduced people's enthusiasm for artificial intelligence. While the public and the industry are still hotly discussing the super capabilities of artificial intelligence in practical applications, the government and academia have begun to pay attention to the innovation it has brought to the scientific research paradigm, and are trying their best to "match" its deep integration with scientific research.

Recently, the Ministry of Science and Technology and the National Natural Science Foundation of China jointly launched a special deployment of artificial intelligence-driven scientific research (AI for Science), which will combine key issues in basic disciplines such as mathematics, physics, chemistry, and astronomy to promote the innovation of artificial intelligence models and algorithms for major scientific problems.

Zhang Xu, the technical director of APUS from the industry, believes that with the continuous breakthroughs in the scale of applications, artificial intelligence has begun to empower all walks of life, including the scientific research field that is not far from the industry. Xu Bo, the leader of the expert group for the implementation of the major project "New Generation Artificial Intelligence" of Science and Technology Innovation 2030 and the director of the Institute of Automation of the Chinese Academy of Sciences, emphasized that the vigorous development of the new generation of artificial intelligence technology is promoting new changes in the scientific research paradigm. "Artificial intelligence has become a new paradigm of scientific research after experiments, theories, and calculations."

The value of enabling scientific research is becoming more prominent

Artificial intelligence technology was born in scientific research and has been developed for more than half a century. In recent years, with the rapid development of deep learning technology and large models, it is natural for artificial intelligence to "feed back" basic research.

Last year, John Jamper led his team to develop AlphaFold 2, which can accurately predict protein structures, and successfully won the 2023 Breakthrough Prize in Life Sciences, known as the "luxury version of the Nobel Prize". The problem of "the amino acid sequence of a protein should be able to completely determine its structure" that has troubled the biological community for more than half a century was finally solved by artificial intelligence. This not only ushered in a new stage in the research of protein structure prediction, but also aroused public attention to AI for Science.

Last year, the Ministry of Science and Technology and six other departments jointly issued the "Guiding Opinions on Accelerating Scenario Innovation and Promoting High-quality Economic Development with High-level Application of Artificial Intelligence", which pointed out that efforts should be made to create several major scenarios and expand the application of artificial intelligence, and high-level scientific research activities are one of them. The launch of the special deployment of AI for Science will undoubtedly further strengthen overall guidance and system layout, give play to my country's advantages in the field of artificial intelligence, and accelerate the paradigm change of scientific research and capacity improvement.

AI for Science is actually about using artificial intelligence to use its powerful data induction and analysis capabilities to learn scientific laws and principles, derive models to solve practical scientific research problems, and especially assist scientists in conducting a large number of repeated verifications and trial and error under different assumptions, thereby greatly accelerating the process of scientific research exploration. Today, this method has achieved remarkable results in many cutting-edge scientific fields.

Shen Dou, executive vice president of Baidu Group and president of Baidu Intelligent Cloud Business Group, believes that artificial intelligence represented by deep learning is expected to become the fourth technological revolution in human history. Its value is more significant than previous technological revolutions in that it not only liberates brain power, but can also partially replace brain power. "This also makes the imagination space of artificial intelligence greater than previous technological revolutions."

In recent years, my country's artificial intelligence technology has developed rapidly, and scientific research data and computing resources have become increasingly abundant. The application scenarios in the field of scientific research have been continuously expanded, which has also laid a solid foundation for accelerating the development of AI for Science. Research in disciplines such as life sciences, mathematics, chemistry, and space science has embraced artificial intelligence.

Zhang Xu said that compared with the AI ​​applications that are familiar to everyone and within reach, the scientific research fields involved in AI for Science, such as biopharmaceuticals, energy, and material research and development, seem far away from people's lives, but the common point behind them is to use AI to "liberate" productivity - to free people from many repetitive and mechanized basic work and conduct more efficient scientific research with the assistance of AI. "These are also the value and charm of AI for scientific research."

Driving the evolution of scientific research paradigms

Ou Weinan, an academician of the Chinese Academy of Sciences and director of the International Center for Machine Learning at Peking University, is full of confidence in artificial intelligence empowering scientific research. He believes that artificial intelligence will greatly improve scientific research efficiency. "AI for Science may push us to be at the forefront of the next round of scientific and technological revolution."

Looking back at the history of scientific development, different stages of development have experienced different scientific research paradigms. Thousands of years ago, humans described natural phenomena through observation and experiment; four or five hundred years ago, the theoretical model paradigm emerged to guide new scientific research; fifty or sixty years ago, after the emergence of large computers, the computing paradigm guided scientific research; in the past 20 years, we have entered the era of big data. Today, artificial intelligence has been able to introduce a new paradigm for scientific research.

In fact, as early as 2021, the technology development trends for 2022 released by Alibaba DAMO Academy pointed out that experimental science and theoretical science have been the two basic paradigms of the scientific community for hundreds of years, and artificial intelligence is giving birth to a new scientific research paradigm.

As an innovator of scientific research paradigm, AI for Science is a process of major reconstruction of disciplines and knowledge. The high threshold requires the collaboration and integration of industry, academia and research. In recent years, many domestic universities and research institutes have actively laid out in the field of scientific intelligence, and domestic enterprises are also actively promoting the development of scientific intelligence and the implementation of the industry.

Of course, as an emerging scientific research paradigm that fully embodies interdisciplinary disciplines, AI for Science involves multiple disciplines and requires a large number of cross-disciplinary scientific research talents. It must also be connected with traditional data set simulation software and data sets in order to gradually form a stable and high-quality scientific research ecosystem.

Ma Yanjun, general manager of Baidu's AI technology ecosystem, told reporters that in promoting the development of AI for Science, enterprises can provide good support in engineering and play an important role in the industrial chain to better drive upstream and downstream industries. "AI for Science requires the cooperation of government, industry, academia, research and application to build a common platform with open source and accelerate vertical integration, which will help scientific research to achieve more results."

Facing the technological trends and emerging scenarios of scientific computing, Ma Yanjun revealed that Baidu has been providing artificial intelligence technologies and platforms for scientific researchers to support and help them complete new explorations in the scientific field. At present, PaddlePaddle is closely integrated with the existing scientific computing ecosystem, actively cooperating with many universities and research institutions to build examples in fluids, materials, biology, etc., and has formed some open, multidisciplinary ecological communities.

From empowering industries to empowering scientific research, the significance of artificial intelligence lies in its accelerating effect on scientific research, allowing more researchers to explore in more complex scenarios, and combine data to infer more accurate physical laws in complex scenarios, and even help them discover new scientific laws.

Looking at the development history of artificial intelligence, we can clearly see its explosion in every stage of development and the surprises it brings to mankind. The current AI for Science is catalyzing a new "technological revolution."