MetaPersona generates realistic personas from social science data
MetaPersona: Task-Grounded Synthetic Populations from Empirical Social Science
Artificial Intelligence
Summary
Creating believable personas for social simulations is hard because it’s unclear which traits to include and how they relate. The authors built MetaPersona-DB, a large collection of social science studies with detailed data about people and their behaviors. Using that, they made MetaPersona, a tool that creates synthetic personas grounded in real evidence, making simulations more accurate. They tested this method on different social topics, seeing good results especially for understanding misinformation beliefs and attitudes towards AI tools. They also made an interactive tool called MetaPersona-Studio to help generate these personas efficiently.
What this means in practice
- •For social simulation developers: Create task-specific synthetic populations grounded in real-world social science data for better behavioral modeling in simulations.
- •For policy analysis teams: Generate realistic synthetic personas to study public opinions and beliefs related to misinformation and AI attitudes in policy scenarios.
Authors
Jinyi Ye, Yuangang Li, Chenxiao Yu, Preyashi Poddar, Priyanka Dey, Longtian Ye, Zihan Wang, Xiyang Hu, Emilio Ferrara, Yue Zhao
Abstract
Personas used to seed LLM social simulations face a cold-start problem: existing methods lack a principled basis for deciding which attributes to include and how to assign their values. As a result, synthetic populations may misrepresent the demographic composition, latent attributes, and dependency structure that shape downstream behavior. We introduce MetaPersona-DB, a dataset of 11,000+ empirical human-subjects studies annotated with task-relevant variables, reported relationships, and aggregate-level population statistics. Building on this resource, we propose MetaPersona, a framework that retrieves task-relevant evidence, constructs literature-derived persona dependency graphs, and samples synthetic populations from empirical priors linking demographics, latent attributes, and outcomes. Across three downstream case studies, three baselines, and three frontier models, results vary by task and model: MetaPersona performs strongly on misinformation belief and AI-tool sentiment, while results on income redistribution are mixed. It also reduces persona-construction cost to under $0.5 per task using GPT-5.2. Finally, we present MetaPersona-Studio, a prototype interactive interface for empirically grounded persona generation.