This page gives some more information about the UKRI AI metascience fellowship project, “Synthetic Metascience: Tracing Artificial Intelligence-generated epistemic shifts In research practice and cultures”, which started October 2025. The project investigates how synthetic data is shaping knowledge-building and epistemic cultures in medical research. You can learn more about the AI Metascience unit here.
Synthetic data offers methods of privacy preservation and data enrichment, particularly important in high-stakes spaces of medical research and practice. I am particularly interested in the following, broad-strokes questions – how does synthetic data mediate epistemic cultures, processes of problem formulation and the nature of scientific evaluation, and reshape notions of data authenticity and truth? I am interested in different types of synthetic data – data generated from specialist ML models for model training, simulation data developed for human learning, and LLM outputs as an emerging form of synthetic data – mapping their overlaps and distinctions in epistemology and epistemic culture.
The aim of this project is to develop insights which feed into AI metascience policy and support building up communities of practice in UK research and practice, contributing to the emerging AI metascience community and literature.
SJ Bennett 