Would You Rely on an Eerie Agent? A Systematic Review of the Impact of the Uncanny Valley Effect on Trust in Human-Agent Interaction

In plain English

A PRISMA review screened 641 records and analysed 53 studies on whether an eerie, almost-human agent is trusted less. Of the 41 that tested the link directly, 17 found a direct effect, 24 found it conditional or indirect, and 10 found none — and trust tracked the agent's usefulness and the quality of its advice more closely than its human-likeness, with perceived competence buffering eeriness. The methodological finding matters more than the headline: most studies used static images or hypothetical scenarios with self-reported trust, only seven used an objective measure, and none measured trust objectively during a real interactive exchange. Practitioners quoting uncanny valley risk should know the evidence base barely tested live conversation.