Shaping the Future of AI-Enabled Surveys
Workshop, June 10th–12th, 2026
AI-enabled surveys use state-of-the-art artificial intelligence (AI) systems to support, augment, or partially automate stages of the survey lifecycle—including questionnaire design, data collection, annotation, and analysis. AI-enabled surveys could help alleviate response burdens and measurement issues, while also raising profound methodological and ethical challenges.
This interdisciplinary workshop is designed to map, advance, and critically assess the emerging domain of AI-enabled surveys. It brings together researchers from the social sciences, survey methodology, natural language processing, and human–AI interaction to articulate a coordinated research agenda. The workshop is structured around three central questions: (1) What are the most effective novel survey modes enabled by AI? (2) How can evaluation standards be reconciled for AI-enabled surveys? (3) How do AI-enabled surveys change the study of society?
By convening an interdisciplinary community around these guiding questions, the workshop aims to delineate the design space of AI-enabled surveys, identify responsible and effective development pathways, and promote consensus on methodological, technical, and ethical best practices. Special focus lies on the definition of core evaluation standards, and on the development of a forward-looking roadmap for future research and implementation in this rapidly evolving field.
Organizers: Georg Ahnert, Indira Sen, Jens Rupprecht, Nicole Schwitter, Ruben Bach, and Markus Strohmaier
Participants: Bernd Weiss, Frauke Kreuter, Max M. Lang, Anna-Carolina Haensch, David Broska, Roberto Cerina, Kristina Gligorić, Fiona Draxler, Claudia Wagner, Jacy Reese Anthis, Sebastian Stier, Dirk Wulff, Tobias Holtdirk, Bolei Ma, Alexandru Cernat, Leah von der Heyde, Jamie Cummins, Dennis Assenmacher, Patrick Sturgis, Florian Keusch, Bella Struminskaya, Jonne Kamphorst, Charlotte Müller, Lisa Steiner, Jessica Daikeler, Barbara Felderer, Maximilian Kreutner, Jana Jung, Abigail Hayes, Leon Fröhling, Bryan Chan, Fridtjof Petersen, Anna Fuchs, Gabriella Lapesa, Simone Ponzetto, Ahmed Salem, Paul Burlon
Presentations
Wednesday, June 10th
- AI in the Survey Lifecycle, Chair: Ruben Bach
- Florian Keusch: What are LLMs used for in survey and public opinion research?
- Anna-Carolina Haensch & Anna Fuchs: AI for Survey Design: Generating and Evaluating Survey Questions with Large Language Models
- Dirk Wulff: Using LLMs to estimate the relatedness of psychology measures
- Patrick Sturgis: SOCbot: Using Large Language Models to measure and classify occupations in surveys
- Leah von der Heyde: How Using LLMs in Survey Research Can Affect Total Error
- Roberto Cerina: Building an Artificially Intelligent Opinion Pollster
- Claudia Wagner: Augmenting Panels: LLM-Based Simulation of Participants
Thursday, June 11th
- Markus Strohmaier: Towards foundation models for opinions & attitudes of the German population
- AI-Augmented Panels, Chair: Markus Strohmaier
- Frauke Kreuter: Adaptive Integrity of AI Models Aligned with Society
- Jens Rupprecht: When Persona Prompting Fails: A Systematic Study of Human Response Variation and Attribute Selection for LLM Adaptation
- Charlotte Müller: Performance of Large Language Models in Creating Synthetic Survey Data
- Jonne Kamphorst: Survey-Based Generative Agents Generalize More Effectively Across Cultural Contexts than Demographic Agents
- Dennis Assenmacher: Imputing Survey Responses with Web Tracking Data: First Insights
- Tobias Holtdirk: In-Context Learning for the Imputation of Public Opinion Data with Large Language Models
- Validity and Evaluation Standards, Chair: Jens Rupprecht
- Kristina Gligorić: Valid Survey Simulations with Limited Human Data
- David Broska: Validating simulations of human behavior
- Indira Sen: Characterizing and Quantifying Errors in LLM-Powered Surveys
- Jamie Cummins: Analytic flexibility threatens the validity of synthetic survey participants
- Alexandru Cernat: Using Multitrait MultiError to estimate data quality
- Georg Ahnert: Evaluating LLM Text Annotations Without Ground Truth
Friday, June 12th
- Open-Endedness and Human-AI Interaction, Chair: Indira Sen
- Bolei Ma: Embracing Open-Endedness in LLM Social Simulation
- Max Lang: No Country for “Old” Surveys – Conversational AI Agents as Interviewers
- Fiona Draxler: AI in Surveys vs. HCI Studies – What we Can Learn from Each Other
- Jacy Reese Anthis: AI-Assisted AI Safety with Human Simulations
We gratefully acknowledge the generous support of the Fritz Thyssen Foundation, the Mannheim Center for Data Science, and the Mannheim Center for European Social Research (MZES), which makes this workshop possible.