Contextualization Reduces AI Persuasion, Warmth Restores It
A study by arXiv (2605.31275) examined how conversational AI personalization affects user trust and reliance. In a 2x2 experiment with 380 participants, researchers tested contextualization (tailoring explanations to user background) and conversational warmth against expert recommendations. Results showed contextualization alone reduced AI persuasiveness, but combining it with warmth restored persuasion via crossover interaction. Reliance on AI remained consistent across conditions, unaffected by conversational design. Trust strongly predicted both persuasion and reliance.
Key facts
- arXiv paper 2605.31275
- 2x2 between-subjects experiment with N=380
- Contextualization reduces AI persuasive power
- Warmth combined with contextualization restores persuasiveness
- Reliance on AI is invariant to conversational design
- Trust strongly predicts persuasion and reliance
- Study focuses on everyday tasks with user lacking prior knowledge
- Personalization identified as persuasive strategy in politics and marketing
Entities
Institutions
- arXiv