A Community Led Risk Register for Responsible AI in Research Software Engineering

Résumé fourni par la source

Adoption of AI tools into research software practices is becoming an increasingly common part of modern RSE practice, yet the community lacks a shared framework for understanding and responding to the risks it introduces. As such, individual practitioners are too often left to navigate the unintended consequences of AI adoption alone, and leaders, institutions and policy makers lack clear visibility of the concerns emerging from RSE practice. Building on an earlier event at Collaborations Workshop 2026, this Birds of a Feather session introduces the Responsible AI Risk Register for Research Software Engineering (RAIRR): a living, community-supported resource that documents risks RSEs have encountered during AI adoption, alongside mitigations, examples of best practice, and assessments of risks’ likelihood, severity, and reach. The register is openly hosted on GitHub and designed for continuous contribution from the RSE community. Working in small groups, participants will discuss the risks most relevant to their own practice, submit new contributions directly to the live register, and comment on other groups' submissions. By the end of the session, attendees will have experienced the full contribution lifecycle and made a meaningful, lasting addition to a public community resource. The session is led by the Software Sustainability Institute's Responsible AI in RSE study group, and serves as an open invitation to join. No prior knowledge is needed - if you use AI tools in your RSE work, or are simply curious about how the community is thinking about responsible adoption, this session is for you.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.