Generative AI based on large language models (LLMs) is rapidly transforming high-performance computing (HPC) and scientific research workflows. This workshop brings together researchers from France and Japan to explore three interconnected frontiers: LLM-driven HPC code generation and autonomous research systems that operate across compiled-language and job-scheduler environments; hardware design automation spanning semiconductor development, HDL generation, and programming for emerging AI processors; and LLM-assisted numerical computation including time-series analysis, floating-point reliability, and linear solvers. Through research presentations and open discussion, participants aim to identify new directions and build opportunities for future international collaborative research.