Introduction Our project, Advancing management skills in AI research, aims to engage the biomedical data science community to identify barriers, document best practices, and develop innovative solutions for managing collaboration and data sharing agreements for AI research. Managing AI research and innovation projects presents new challenges for research professionals, necessitating both innovation in practices and workforce upskilling. Key barriers to progress include difficulties in cross-sector collaboration and data sharing, both essential for AI development. Formal collaborations require legal agreements, which are often bespoke and resource-intensive, slowing research progress. Data sharing poses further challenges, especially in AI, where large, diverse datasets are needed for model training. Existing data access procedures are typically inadequate for such needs, and transferring data to centralized repositories is often restricted by regulations, necessitating custom solutions that consume time and resources. Event details This event was shaped by an expert working group who helped us define the core challenges and focus areas, and who are also contributing to the event planning and execution. We aim to bring together stakeholders from various sectors and disciplines, including AI researchers, biomedical data scientists, research managers, legal and finance professionals, and knowledge exchange experts. Beyond knowledge sharing, the event will facilitate cross-sector collaboration to surface and refine innovative, actionable practices that can have a real, immediate impact in the field. The event took place in London on Wednesday 04 February 2026. After the event After the event, we will produce a final project report, including literature findings, event outcomes, and case studies. This report will offer recommendations and be published openly, with all materials archived on the Zenodo repository under open licenses to support reuse and sustainability. All contributors to this project will be acknowledged in the report and any other project outputs.
Karoune et al. (Thu,) studied this question.