AI4S’s eight-year accumulation has ushered in a new node, and Beijing has fallen behind again, taking the lead in the second half of the transformation of the scientific research paradigm.

Zhixixi Author | Chen Junda Editor | Mo Ying Zhixixi reported on August 22 that yesterday, the 2026 Scientific Intelligence Conference opened at the Zhongguancun International Innovation Center in Beijing. This conference gathered nearly 50 academicians from the two academies and more than 80 young experts, set up 2 plenary sessions and 14 parallel sessions, and held hundreds of academic reports. About 2,500 people from governments, universities, research institutes and enterprises attended. Judging from the agenda, the topics have expanded from models and methods to scientific data, independent laboratories, pilot verification and industrial implementation; the progress of the country's first independent laboratory national standard project was also disclosed at the meeting. During the conference, there were two industry trends that deserved special attention. One is that the Beijing Municipal Science and Technology Commission and the Zhongguancun Administrative Committee released the first batch of 16 typical cases of artificial intelligence empowered scientific research; the other is that Haidian District officially released the Zhongguancun (Haidian) Scientific Intelligence (AI4S) Innovation Cluster, which is located in the Xisanqi area and coordinates 1.6 million square meters of industrial space and 740,000 square meters of pilot space. Among the 16 typical cases, 12 are located in Haidian, and the cluster area is also located here. Going back further, the concept of AI4S, the world's first new R&D institution focusing on AI for Science, and the emergence of the first batch of companies are all in this district. Behind this concentration is Haidian’s continuous accumulation in AI4S over the past few years. How this accumulation is formed must start from the source of the concept. 1. Starting from a concept 8 years ago, AI4S has entered the second half. The time goes back to around 2018. At that time, E Weinan, an academician of the Chinese Academy of Sciences and a chair professor at the School of Mathematical Sciences at Peking University, began to systematically promote the concept of AI for Science and advocated using AI to reshape the scientific research paradigm. He not only proposed ideas, but also produced representative results. The deep potential energy method developed by E Weinan uses machine learning to fit atomic interactions with first-principles accuracy, which simultaneously improves the scale and accuracy of molecular dynamics simulations by several orders of magnitude. At that time, this was only the direction of a few people, and the mainstream AI community's attention was still on images, speech, and later large models. ▲Eweinan The turning point occurred at 2