What can we do about it?

The aims of this workshop were to: 

• Co-create practical and meaningful solutions to the challenges identified in the previous workshops, to promote data transparency, quality and integrity  

• Consider when and how the reliability of data-based applications can be scrutinised before they are used in decision-making, e.g. models and AI in patient treatment 

Session one: Making Data FAIR

Sofia Fernandes, Associate Director and Head of Research and Academic Services, University of Warwick, introduced the principles of Findability, Accessibility, Interoperability and Reusability of research data (the FAIR principles). Sofia explained why the principles are a necessary condition for meaningful transparency, how they help create opportunities for scrutiny, and how they can help enhance societal benefits of research. 

Session two: Scrutinising Data Quality

Fiona Booth, Director, Data Integrity and Quality Assurance programme, University of Bristol, discussed the processes and practices that improve data integrity and quality in the public interest. Data is not necessarily high-quality even if it is FAIR. What does someone in her role do to check or interrogate the quality of data? 

Session three: Asking the Right Questions About Reliability

Jennifer Ding, Senior Researcher in Research Applications, Alan Turing Institute, introduced the processes, practices and questions that can help improve the reliability of data-based applications in society (e.g. Artificial Intelligence in patient treatment, or climate simulation models). What do we need to know about the data used to train a model or AI? What do we need to know about how bias has been mitigated at each stage? What do we need to know about how a model works and the assumptions it makes? What do we need to know about the capabilities of an AI in the real world before it is widely rolled out? What is the existing role of researchers, research institutions, developpers and policymakers in scrutinising reliability? Jennifer introduced the work of the Alan Turing Institute in promoting and supporting scrutiny of research data applications.