AI
MetaPersona: Task-Grounded Synthetic Populations from Empirical Social Science
Publication: cs.AI updates on arXiv.org
Author: Jinyi Ye, Yuangang Li, Chenxiao Yu, Preyashi Poddar, Priyanka Dey, Longtian Ye, Zihan Wang, Xiyang Hu, Emilio Ferrara, Yue Zhao
Source published: 01/10/2026, 04:00:00 UTC
Briefing created: 01/10/2026, 04:12:17 UTC
Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 01/10/2026, 04:03:25 UTC.
Briefings
MetaPersona-DB provides 11,000+ annotated empirical studies to address cold-start problems in LLM synthetic populations. MetaPersona framework constructs dependency graphs from literature and samples from empirical priors, reducing construction costs to under $0.5 per task. Results show strong performance on misinformation belief and AI-tool sentiment, but mixed outcomes on income redistribution.
Key points
- MetaPersona-DB contains 11,000+ empirical human-subjects studies annotated with task-relevant variables, reported relationships, and aggregate-level population statistics.
- MetaPersona framework retrieves task-relevant evidence, constructs literature-derived persona dependency graphs, and samples synthetic populations from empirical priors linking demographics, latent attributes, and outcomes.
Context and limitations
Results vary by task and model across three case studies and three frontier models; specific performance metrics for income redistribution remain mixed.
Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 01/10/2026, 04:12:17 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.
View archived briefing on IPFS
bafybeibgaqdyeg36mgb4kus5sfpkteuf3lb6fsb4zie5nashtv7sxgicj4