The ELSA Way: A community-driven knowledge platform for responsible medical AI

Issue

Responsible AI development in healthcare requires integrating ethical, legal, and societal (ELSA) considerations throughout the AI lifecycle. Across research, ample expertise has been built, yet much remains tacit, held in people and projects, locked in publications, or expressed in language inaccessible to developers, policymakers, and communities. No shared, practical infrastructure exists to capture and disseminate this knowledge.

Description

We propose a protocol for The ELSA Way, a freely accessible knowledge platform inspired by The Turing Way. The protocol describes the design, content scope, and governance model of The ELSA Way, and integrates research conducted with AI developers, healthcare professionals, and patients (including surveys, interviews, and co-creation workshops) to ground content development.

Objectives

1) make explicit and disseminate tacit ELSA knowledge; 2) translate research findings into practical guidance; 3) create extensible infrastructure applicable across responsible AI initiatives and domains.

Results

Initial content grounded in ELSA research is under development, including frameworks, case studies, and worked examples. User and contributor needs, design priorities, and governance principles have been identified through an inquiry. The community-driven model generates added value beyond any single lab’s capacity: aggregating cross-domain ELSA expertise, reducing duplication of efforts, and creating a shared language for responsible AI based in diverse real-world contexts.

Lessons

Tacit knowledge externalisation requires dedicated infrastructures. Open science tools lower barriers to contributions. Community contribution and governance are essential for quality and sustainability. These lessons are embedded in the protocol, which is intended to be transferable to any AI domain where ELSA expertise risks remaining inaccessible, and is designed to be adopted, adapted, or co-developed by the research community and society.

Key messages

• Knowledge on responsible AI in healthcare remains largely tacit and inaccessible. A practical knowledge infrastructure is needed to make ELSA knowledge explicit and usable beyond academia.

• The ELSA Way is an open, community-driven platform bringing together knowledge and translating research findings into accessible guidance for responsible AI development across sectors and countries.