Through the Looking-Glass: AI-Mediated Video Communication Reduces Interpersonal Trust and Confidence in Judgments

AI-mediated video tools reduce perceived trust even when they don't affect accuracy: The study found that participants reported lower trust and confidence in their judgments when watching AI-enhanced videos (with retouching, background replacement, or avatars), yet their actual ability to distinguish truthful from deceptive statements remained unchanged across all conditions.

Inconsistent use of AI tools creates the biggest trust problem: Trust declined most sharply "in settings in which some participants used avatars while others did not," suggesting that mixed adoption of AI mediation tools—rather than the technology itself—may be the primary concern for interpersonal communication.

People don't automatically assume AI users are lying: Despite reduced confidence, participants were "no more inclined to suspect those using AI tools of lying," indicating that concerns about AI-mediated communication triggering false suspicion of deception may be overstated.

The trust-accuracy gap matters for real-world applications: While AI mediation didn't impair lie detection accuracy, the decline in perceived trust and confidence has practical implications for contexts "where not only truth, but also trust and confidence matter"—such as job interviews, legal proceedings, or sensitive negotiations.

Design choices in AI tools deserve attention: The findings emphasize "the importance of trustworthy AI mediation tools," suggesting that how these technologies are implemented and deployed matters as much as whether they're used at all.