AI, Community Data and Public Decision-Making: Turning Lived Experience into Trustworthy Evidence

About the Project

Public decision-making is increasingly shaped by data. Yet the evidence used by health systems, local authorities, and other public bodies is still dominated by administrative data designed for performance monitoring, compliance, and cost control. This creates a major blind spot. Such data can show volumes, waiting times, or service outputs, but often fails to capture the lived experience, local knowledge, and social context that shape need, access, and outcomes.

At the same time, voluntary, community and social enterprise organisations generate rich forms of community data through their everyday work. This may include case records, referral notes, reports, feedback, and other forms of qualitative and semi-structured information. These materials often provide insight into unmet need, barriers to access, and the realities of people’s lives in ways that administrative systems cannot. However, this data is typically fragmented, inconsistent, and stored in forms that are difficult to analyse or integrate into formal decision-making processes.

This PhD will explore how artificial intelligence can help address that problem. It will investigate how methods such as natural language processing, information extraction, knowledge graphs, or privacy-preserving machine learning might be used to transform unstructured community data into structured, reusable, and policy-relevant evidence. The project will not be concerned only with technical performance. It will also examine questions of trust, representation, governance, ethics, and accountability. In other words, the PhD will ask not only whether AI can make community data usable, but under what conditions it can do so responsibly and without stripping out the social meaning that gives that data value in the first place.

The PhD is likely to combine methodological development with one or two applied case studies in areas such as health and social care, community wellbeing, or wider public service delivery. It may involve close collaboration with VCSE organisations and public-sector partners to understand real data environments, co-design analytical approaches, and assess how resulting tools or pipelines perform in practice. The candidate would therefore work at the intersection of AI, public policy, data governance, and community-based research.

The project would make three main contributions. First, it would contribute to knowledge on how AI can be adapted to messy, real-world community data rather than only clean and standardised datasets. Second, it would advance understanding of how lived experience and community knowledge can be translated into forms that are useful for policy and commissioning while remaining ethically grounded. Third, it would generate practical lessons for the design of trustworthy, community-informed data infrastructures for public decision-making.

Supervisors

Daniel King

Juliana Mainard-Sardon