When algorithms evaluate people: how AI video interviews relate to employability confidence and organizational...

Algorithmic evaluation is increasingly entering high-stakes social contexts, including recruitment, where young people are assessed before any human interaction occurs. AI video interviews are one such setting; they are largely one-directional, opaque, and often provide no feedback about how applicants are evaluated. This study examines how such AI-mediated evaluation relates to two socially consequential perceptions among university students: confidence in their own employability and trust-related evaluations of organizations that deploy AI-based assessment. Drawing on Situated Expectancy-Value Theory (SEVT), we distinguish between expectancy beliefs-whether one can succeed-and value beliefs-whether the outcome is worth pursuing, and develop a dual-pathway model in which expectancy-related individual factors (AI literacy, metacognition, need for cognition, and technology anxiety) are associated with perceived employability, while value-related process perceptions (procedural justice and algorithmic transparency) are associated with organizational attractiveness, with social comparison orientation and AI usage experience as boundary conditions. Using data from 520 Chinese university students analyzed with PLS-SEM, we find that all four expectancy-related factors are independently associated with perceived employability, and that procedural justice and algorithmic transparency are each substantially associated with organizational attractiveness. AI usage experience and social comparison orientation further moderate the hypothesized relationships. The findings contribute to debates on AI-mediated evaluation, algorithmic transparency, and the social consequences of using AI in early career transitions.