Episode Details

Most job interviews are broken — and the science to prove it has existed for three decades. The Schmidt and Hunter meta-analysis found that a typical unstructured interview, where a hiring manager just asks whatever comes to mind, has a predictive validity of .38. Structured interviews, where every candidate gets the same questions scored against consistent anchors, hit .51 — a 34% improvement in predicting who will actually succeed on the job. IO psychologists identified 15 specific components that drive that improvement: anchored scoring rubrics, job-analysis-based questions, no improvised follow-ups. The problem has always been delivering all 15 consistently at scale. Most organizations manage a handful at best. That's what asynchronous AI video interview platforms like OVI are built to solve. Every candidate gets identical questions in the same order. No rapport bias, no halo effect, no similarity attraction pulling scores off course. AI-scored async interviews achieve criterion-related validity around .24 — squarely within the validated range — with meaningfully less adverse impact against underrepresented candidates. OVI takes an extra step with transcript-only analysis: no facial recognition, no emotion detection from voice, just what candidates actually say. It's the human-in-the-loop approach IO psychology's governing body recommends, starting at $99/month.