Research computing and data (RCD) infrastructure increasingly operates as a global enterprise. Researchers, support staff, and administrators collaborate across dozens of languages and cultural contexts, yet the tools and practices for managing multilingual communication in cyberinfrastructure (CI) environments have received little systematic attention from the community. At the same time, rapid advances in large language models (LLMs) and AI-powered translation are transforming what is technically possible: recent benchmarking shows LLMs now approach or exceed traditional machine translation systems across many language pairs and newer studies demonstrate that LLMs can preserve scientific meaning when translating research articles across 28 languages; A comprehensive survey of multilingual LLM architectures further highlights both the rapid progress and the remaining gaps, particularly for low-resource languages. This presentation will introduce the current landscape of these multilingual AI technologies and examine their potential applications across RCD operations, such as translating HPC technical documentation and support tickets and making cybersecurity advisories and incident reports accessible across language barriers. The presentation will also call for practical considerations and discussions, including translation accuracy for specialized CI terminology, data privacy when routing sensitive communications through AI services, and the limitations of current tools.
Pengyin Shan (Wed,) studied this question.