Meeting the AI and Big Data Revolutions in Quantitative Social Science

Introducing the LIDA: Science of Data Science Programme. Join us for lectures and interactive discussion followed by drinks and networking in NEXUS.

Quantitative social science stands at a critical crossroads. The twin revolutions of big data and AI are transforming our research landscape, bringing a dizzying mix of new opportunities and tricky methodological, ethical, and epistemological challenges.

This inaugural session, hosted jointly by the Science for Data Science Programme and the Social Research Methods Centre at Leeds, will examine how these technological disruptions are reshaping the science of social inquiry. How can scientists harness these innovations while maintaining their rigor? What new questions can be answered? And what established practices may require reconsidering?

The event features three thought-provoking presentations from the cutting-edge of quantitative social science research, followed by a facilitated discussion on implications for research practice and disciplinary evolution. Join us in this critical conversation about navigating the future of quantitative social science.

Speakers include

• Dr Peter Tennant
• Prof Daniel Birks
• Prof Mandeep Dhami

Talk 1 - AI in Quantitative Social Science Research: Promises and Perils
(Dr Peter Tennant)

Abstract:

Modern machine learning algorithms, typically branded as ‘artificial intelligence’, promise to revolutionise the science and practice of quantitative social science research. Many routine or analytical tasks could potentially be improved, or replaced, by AI. These include improved prediction modelling, literature searching, preparing and scrutinising code, extracting and synthesising information, and even conducting quantitative analyses. Unfortunately, it is difficult to separate the genuine promise from the hype and truly appraise the current capabilities and pitfalls. This talk will critically examine the potential and the perils of using current and near-future AI technology in quantitative social science research.

Bio:

Peter Tennant as an Associate Professor of Health Data Science at the University of Leeds, former fellow of The Alan Turing Institute, and the incoming John S Saden Visiting Associate Professor at Yale University in 2026. He is renowned for his work on translating causal inference theory and methods into applied health and social science research. He leads the Science of Data Science Programme at the Leeds Institute for Data Analytics and the Causal Inference Interest Group at The Alan Turing Institute.

Artificial Intelligence Powered Tools: The Future of Social Science Research
(Prof Mandeep Dhami)

Abstract: JuDDGES (Judicial Decision Data Gathering, Encoding and Sharing) represents an example of the type of game-changing tool that will be available to researchers in the future (see https://juddges-project.eu/). The two-year project received funding from a CHIST-ERA network of research funders: EPSRC (UK), NCN (Poland) and ANR (France) and involves a multi-disciplinary and multi-national team of researchers. Development of the JuDDGES tool is squarely shaped by the needs of researchers, and applies Open Science principles. The tool capitalizes on Artificial Intelligence (AI), Machine Learning (ML), Natural Language Processing (NLP) and Human-in-the-Loop technologies. In this talk, I will discuss the need for such a tool, describe the aims of the project and progress to-date, as well as the anticipated impact.

Bio: Mandeep K Dhami, PhD, is Professor of Psychology at Middlesex University, London, UK. Mandeep is an internationally recognized expert on human judgment and decision-making, risk perception, and uncertainty communication. She applies her expertise to solving problems in the criminal justice and defence & security sectors. Mandeep has authored over 140 scholarly publications and is lead editor of the book ‘Judgment and Decision Making as a Skill’ published by Cambridge University Press. Her research has received several international awards including from the NATO Science & Technology Organisation. She regularly advises Government bodies nationally and internationally on evidence-based policy and practice. Mandeep is currently also Governor of Middlesex University London, Editor of Judgment and Decision Making, and President-elect of the European Association for Decision Making.

Using Large Language Models to Identify Indicators of Vulnerability in Police Incident Narratives
(Prof Daniel Birks)

Abstract:

This talk presents recent research from the ESRC Vulnerability and Policing Futures Research Centre, exploring if instruction-tuned large language models (LLMs) can support qualitative analysis in criminological research. Police officers and dispatch handlers routinely record unstructured narrative data describing police–civilian interactions. These narratives offer valuable insights into engagements with vulnerable populations and can inform evidence-based problem and demand analyses, officer training, and inter-agency coordination. However, manual coding of such data is often prohibitively resource-intensive.

Using publicly available narrative reports from the Boston Police Department, we assess if LLMs can effectively replicate human qualitative coding of four key vulnerabilities: mental ill health, substance misuse, alcohol dependence, and homelessness. Human-generated classifications are compared with those produced by various LLMs using different prompting strategies. Counterfactual experiments are also conducted to test for potential classification biases related to subject race and sex.

The talk outlines the results of these analyses and considers where and when deployment of LLMs might enhance the capacity to analyse large-scale unstructured datasets, with implications for both evidence-based policing and social science research more broadly.

Bio:

Dan is Professor of Computational Social Science at the University of Leeds School of Law. His research focuses on the responsible use of administrative data to inform evidence-based policy and practice. He currently serves as Deputy Director of the ESRC Vulnerability & Policing Futures Research Centre, where he leads research into how linked administrative data can support public service responses to vulnerability.