The research team led by Marcus J. Buehler from the Massachusetts Institute of Technology has unveiled the 'Recursive Meta-Intelligence' (RMI) research paradigm. This innovative paradigm empowers AI to autonomously create a comprehensive toolchain for the mechanical analysis and simulation of metamaterials. This encompasses mechanical solvers, geometric generators, and more, all generated using only natural language descriptions and reference images. Furthermore, it can output 3D printing structure files. Leveraging this tool, the AI agent, after conducting 6,000 simulations, overturned the long-held consensus in the metamaterials community that 'a greater number of hierarchical levels results in superior damage tolerance.' Instead, it discovered that fracture behavior is predominantly determined by the spatial arrangement of material distribution and geometric orderliness. The related research findings have been made publicly available as open-source resources. At present, although this paradigm represents an early stage of 'scientific superintelligence,' enabling AI to handle the entire mechanical research process, the accuracy of the AI model still falls short of that of commercial software, and the conclusions drawn still necessitate physical verification. Additionally, the active feedback and continuous improvement of the tools by AI have not yet been realized.