Talking to Machines

Large Language Models workshop

Wednesday, 25th June
Carlos III University – Madrid, Spain

Livia Schubiger (D-GESS, ETH Zurich) & Raymond Duch (University of Oxford) are organizing an EPSA Madrid 2025 pre-conference that will assemble researchers working with Large Language Models (LLMs). The focus will be on the application of LLMs to research being conducted in the social sciences. The broad themes that we expect to cover include:

  • Data generation: this includes the role of AI agents as subjects in experiments, as enumerators, or as confederates; adaptive experimental design.
  • Data collection and curation:  examples might include the compilation of historical and contemporary records videos and text.
  • Data analysis:  text analysis
    and classification; the analysis of treatment effects associated with text and image vignette; the role of LLMs in the analysis of conversations and open-ended survey responses.
  • Design:  LLM guidance in the design of experimental protocols and treatment arms.
  • Technical advances including the incorporation of non-text data such as images, video and audio.
  • Safety, ethics and replication:  the ethical challenges of incorporating LLMs into social science research; efforts to address safety and LLM model development and deployment.
  • LLMs, the economy and governance: understanding how the rapid introduction and adoption of LLMs will affect the economy and governance.

We are at an early stage of incorporating LLM applications into our social science research.  Hence this European event is a great opportunity to showcase the innovative research that is being conducted by scholars world-wide.  The workshop format offers presenters a unique opportunity to showcase their work and get constructive and thoughtful feedback.

TimingRoomPresentation Schedule
Registration8:30Foyer just outside the Auditorium
Welcome and Introduction09:00 – 09:10AuditoriumLivia & Ray
Panel Session 109:10 – 11:00AuditoriumDiscussant – Frederik Hjorth, University of CopenhagenArianna Muti, Bocconi University (Milan) “Where are the leftist feminists now?” An LLM approach to how political events influence social media discourse on feminism
Christy Coulson, London School of Economics and Political Science (LSE) “Gender-based Activism and Violent Backlash: Evidence from Latin America”
Matilde Ceron, University of Salzburg “Leveraging Crowdcoding and LLMs for the analysis of gender mainstreaming: gender-sensitive recovery policies in the EU”
Nils-Christian Bormann and Edoardo Alberto Viganò, Witten/Herdecke University. “Historical Conflict Event Data Collection via Large Language Models (LLMs)”
Ashrakat Elshehawy, Stanford University “How Biased Police Reporting Shapes Misperceptions of Out-Group Crime”
Break11:00 – 11:20Foyer just outside the Auditorium
Panel Session 211:20 – 13:00AuditoriumDiscussant – Elizabeth Rhodes, OpenResearchGloria Gennaro, KCL “The evolution of political rhetoric on immigration in the UK”
Sascha Riaz, European University Institute “Regime Loyalty during Wartime – Evidence from Nazi Germany”
Tore Wig, University of Oslo “Prompting for Theoretical Progress: Using Generative AI to evaluate the evidence support for Grand Theories of Politics” 
Xiao Liu, Peking University “Automated Annotation of Evolving Corpora for Augmenting Longitudinal Network Data: A Framework Integrating Large Language Models and Expert Knowledge”
Christopher Klamm, University of Cologne “Measuring Personal Attacks in Parliamentary Debates”
Workshop Lunch 13:00 – 13:30 Foyer just outside the Auditorium
Panel Session 313:30 – 15:10AuditoriumDiscussant – Xun PANG, Peking UniversityJavier Osorio, University of Arizona “ConfliBERT: A Language Model for Political Conflict” 
Jongwoo Jeong, Georgia State University “Spelling correction with large language models to reduce measurement error in open-ended survey responses”
Winnie Xia, Aarhus University, Yen-Chieh Liao & Slava Jankin,  University of Birmingham “Dynamic and Multilingual Embedding Regression for Political Text”
David Muchlinski, Georgia Tech “SCOPE: Supercharging Conflict Prediction and Explanability using Advanced AI”
Giovanni Pagano and Luigi Curini University of Milan “More than words? Understanding Multimodal Political Communication: a Computational Analysis of Textual and Visual Elements in the EP 2024 Campaign”
Break15:10 – 15:20Foyer just outside the Auditorium
Panel Session 415:20 – 17:00AuditoriumDiscussant – Moritz Marbach, UCLGiuliano Formisano, University of Oxford “The Network Dynamics of Online Polarisation in the US”
Camilo Cristancho, Universitat de Barcelona “Attitudes towards protest: A large-scale comparative perspective of media representations of protest in Latin America 2000-2024”
Rachel Bernhard, Nuffield College, University of Oxford “Kiss, Marry, Kill: Appearance-Based Discrimination in Politics”
Bryce J Dietrich, Purdue University, Using AI to Summarize US Presidential Campaign TV Advertisement Videos, 1952–2012
Taehee Kim, University of Konstanz “Evaluating Generative Agents in Social Media: Replicating User Attitudes and Behaviors”
Panel Session 517:10 – 18.40AuditoriumDiscussant – Isaac Mehlhaff, Texas A&M University Bastián González-Bustamante, Leiden University “Charting Reproducibility and Performance: LLMs in Multilingual Toxic Speech Detection”
Fabrizio Gilardi, University of Zurich “Understanding and Mitigating Online Toxicity: A Chatroom-Based Experimental Framework with LLM Agents”
Linette Lim, University College Dublin, Yen-Chieh Liao & Slava Jankin,  University of Birmingham  “Who Believes and Who Shares Fake News: A Multi-Agent System Application for Experimental Misinformation Research with LLMs”
Nicolai Berk, ETH Zurich “What is a Good Conversation? Improving the Operationalization and Measurement of Deliberative Quality by Analyzing Interactions”
Moritz Osnabrügge, Durham University “Polarized Speech: How Elite Rhetoric and Echo Chambers Fuel Negative Emotive Political Debate”

Papers