AI Quest

Large Language Models and other AI tools are increasingly being used in the field of evidence synthesis. Whilst approaches for using AI to assist with quantitative reviews are now well-established, the area of qualitative evidence synthesis has been relatively overlooked, which is curious given the unique power of LLMs in abstract thematic reasoning. Contemporary models are opaque and prone to hallucination. We are working with the WHO to use best qualitative research practices to build a model from scratch that can perform descriptive thematic analysis on sets of policy documents to answer questions like ‘which policy approaches work best in [context x]?'

Bradley Segal is leading the software development, with Luke Allen coordinating the wider team that draws from experts across the department and wider university. If you are interested in our work