"Astronomy is one of the most open disciplines, and it is a cutting-edge science that breeds major original discoveries." From August 4 to 7, the second open science conference on astronomy enabled by artificial intelligence was held in Qinghai. Zhou Qi, an academician of the CAS Member, vice president of the Chinese Academy of Sciences, and president of the University of the Chinese Academy of Sciences, said that at present, artificial intelligence (AI) is bringing profound changes to scientific research, and the astronomy field urgently needs to use AI to accurately decode the universe. At the meeting, leaders, experts, and young researchers from nearly 30 astronomical research institutions in China, the United States, the United Kingdom, France, and other countries exchanged and discussed the main theme of the deep integration of artificial intelligence and modern astronomy.
"The Chinese Academy of Sciences will continue to rely on the Cold Lake Astronomical Observation and Research Base in Haixi Prefecture to focus on major scientific issues such as dark matter, dark energy, the origin and evolution of the universe, promote the deep integration of AI with observation and theory, and release data value. ”Zhou Qi stated that she will deepen international cooperation and technological governance, and contribute wisdom and strength to the development of global astronomy with a more open and inclusive mind.
With the rapid development of large-scale ground-based sky surveys, space exploration missions, and multi messenger astronomy, modern astronomy is entering a new stage of data intensive, time sensitive, and cross domain collaboration. The new generation of astronomical observation facilities continues to generate massive, multi-source, multi band, and multi messenger observation data, and the data scale has rapidly moved from TB (terabyte) level to PB (petabyte) level and even EB (exabyte) level. The traditional data processing mode that relies on manual analysis and a single algorithm is no longer able to meet the needs of scientific research.
"A first-class site plus first-class equipment is not equal to first-class scientific output." Liu Jifeng, director of the National Astronomical Observatory of the Chinese Academy of Sciences, said that after the telescope is built, it is the real challenge to make it efficient, intelligent and sustainable to produce high-quality data. And the key solution to this problem is AI, "he said.
Hu Yi, a researcher at the National Astronomical Observatory of the Chinese Academy of Sciences, AI、 The development of new generation intelligent technologies such as big models is expected to provide new technological paths for efficient processing, knowledge mining, and scientific discovery of massive scientific data, and can also promote the gradual evolution of astronomical observations from "data acquisition" to "intelligent analysis and autonomous discovery".
Currently, the construction of a "smart observatory" has become a key development direction of concern in the international astronomical community. It takes AI as the core driving force to build an intelligent closed-loop system covering the entire process of observation, operation, and processing, promoting the evolution of the observatory from traditional observation facilities to a new generation of intelligent scientific platform that integrates intelligent operation, intelligent observation, intelligent analysis, intelligent decision-making, and independent scientific discovery.
Hu Yi further introduced that the intelligence level of the smart observatory can be divided into five levels, from the auxiliary artificial level of L1 to the autonomous exploration level of L5. The intelligence level, task autonomy, and scientific contribution ability of the telescope system gradually improve with the increase of levels.
During the conference, the attending experts also conducted exchanges and discussions on the basic astronomical model OneAstronomy, as well as China's independently developed artificial intelligence astronomical models such as StarWhisper, Jinwu Sun Large Model, and SpecCLIP Spectral Model.
Among them, OneAstronomy was jointly developed by Zhijiang Laboratory and the National Astronomical Observatory of the Chinese Academy of Sciences. It focuses on four types of core observational data: astronomical images, spectra, time-domain data and catalog data, explores the use of artificial intelligence to achieve unified representation, fusion analysis and knowledge discovery of multimodal astronomical data, and helps researchers identify important targets and discover potential laws from the growing mass of observational data.
Researcher Shi Junhui from Zhijiang Laboratory introduced that OneAstronomy will provide artificial intelligence technology support for major scientific projects such as the Chinese Space Station Survey Telescope and the Global Open Source Telescope Array, and serve the global astronomical research community through open cooperation. With the continuous improvement of telescope observation capabilities, the amount of data generated is rapidly increasing. Artificial intelligence not only needs to understand human language, but also needs to be able to understand scientific data such as images, spectra, and light curves, helping scientists discover noteworthy clues from massive observations faster, "he said. (Looking into the New Era)