Pew study: AI synthetic survey respondents can't stand in for humans
IanArawjo · x · 2026-11-16
Pew Research Center ran a methodological experiment having AI models role-play real respondents from its American Trends Panel and answer the same surveys. Verdict: synthetic surveys do a poor job of estimating public mood, with especially large errors for some subgroups like Republicans and Black adults, and results shift depending on the model. Pew has no plans to use AI-generated survey data.
More from Research
- ICLR Reviewer: DRL Papers Are Top-Tier, LLM Work 'Riddled with Experiment Confounds' — tw_killian · 2026-11-16
- Artificial Humanities lab's year in review: papers in 2026, books in 2027 — begusgasper · 2026-11-16
- ASI as a Black Box: we can deduce almost nothing about what superintelligence could do — zetalyrae · 2026-11-16
- wordgrammer: a good benchmark must be a task you don't care about — wordgrammer · 2026-11-16
- Turing Post Guides: 15+ Forcing Methods for Video Generation and 10 Looped Transformer Papers — TheTuringPost · 2026-11-16
- 156-page open book on knowledge graphs: RDF, OWL, SHACL, Neo4j and Graph RAG — adnan_hashmi · 2026-11-16