At the 2026 World Artificial Intelligence Conference (WAIC) and the High-Level Meeting on Global AI Governance held in Shanghai from July 17 to 20, a dedicated forum titled "AI and Advanced Nuclear Energy Co-building a Sustainable Intelligent Future" was organized by the Institute of Modern Physics, Chinese Academy of Sciences.
The forum featured the launch of the world's first AI for ADANES technical roadmap, the establishment of the inaugural All-domain Innovation Alliance for AI for ADANES, and the release of an industry consensus comprising eight core propositions.
At the event, Academician Zhan Wenlong of the Chinese Academy of Sciences, the Chief Commander of the ADANES project, and He Yuan, the Chief Engineer of CiADS and the Chief Technologist for ADANES, alongside representatives from universities and enterprises, jointly unveiled this pioneering technical pathway.
This roadmap establishes an intelligent system based on a physical intrinsic world model, introducing a novel paradigm driven by the synergy of data, physical models, and expert experience. It addresses the fundamental conflict between the "uncontrollable black box" nature of AI and the "zero-tolerance for error" safety imperative of the nuclear industry, thereby filling a global technological gap in the intelligent implementation of advanced nuclear energy.
ADANES (Accelerator Driven Advanced Nuclear Energy System), reported to be China's sole original solution in the field of advanced nuclear energy, was originally proposed by the Chinese Academy of Sciences. It integrates nuclear fuel breeding, spent fuel transmutation, and power generation. While its subcritical operation alters the safety logic of traditional reactors, it also introduces new challenges related to the strong coupling of accelerator, target, and reactor systems and ensuring long-term stable operation.
The newly released native AI technical pathway systematically proposes a five-layer architecture encompassing a unified data source, a physical intrinsic world model, controllability of the physical world, collaborative scheduling of intelligent agents, and continuous learning and evolution. This framework aims to embed artificial intelligence natively throughout the entire lifecycle of ADANES, including design, commissioning, operation, and maintenance.
The concurrently established AI for ADANES Alliance brings together research institutes, nuclear power enterprises from both fission and fusion sectors, AI companies, financial institutions, and other stakeholders. The alliance intends to leverage large-scale scientific research facilities as a platform for the co-creation and sharing of comprehensive intelligent solutions.
Addressing the boundaries of AI in nuclear safety, He Yuan stated that AI can evolve from a "child prone to fantasy" into an "expert," but this requires long-term cultivation by the physical world and human experts. He emphasized that the final decision-making authority for nuclear safety must remain in human hands.
In the closing session, He Yuan announced the forum's consensus, outlining eight core propositions: clarifying the strategic positioning of mutual empowerment between AI and nuclear energy, establishing the triple-drive technical paradigm, adhering to the zero-tolerance bottom line for nuclear safety, opening the CiADS large-scale facility as a shared platform, pioneering the new track of "advanced nuclear energy + zero-carbon computing power," strengthening the foundation of interdisciplinary talent, deepening international exchange and governance collaboration, and jointly exploring a safe and inclusive path for intelligent energy.
Comments