AI Interview Tools: What the Research Actually Says About Accuracy and Bias

HireVue processes more than 40 million interviews a year, and nearly a third of large enterprises now use some form of AI-powered interview scoring. But what does the independent research actually say about whether these tools work — and whether they're fair? In this episode, we dig into the growing gap between vendor promises and peer-reviewed evidence. Independent studies consistently place AI interview scoring tools at validity coefficients of .20 to .35 — comparable to an unstructured phone screen, not the structured behavioral interviews they're marketed as replacing. And the bias findings are harder to dismiss: a 2023 University of Melbourne study found statistically significant bias against non-native English speakers across all six AI hiring tools it tested. Yet only 22 percent of employers actively test their tools for bias, according to SHRM's 2024 survey. Candidates are taking notice too. LinkedIn's 2024 Global Talent Trends report found that 38 percent of candidates would turn down a job offer from a company that used AI screening without disclosing it. That's not an abstract ethics issue — it's a conversion rate problem. We also walk through the regulatory pressure building from every direction: EEOC guidance placing liability squarely on employers, NYC Local Law 144 requiring annual independent bias audits, FTC enforcement signals, and the EU AI Act's high-risk classification for employment AI. The window for compliance prep is closing fast. We close with five questions every HR leader should ask before adopting — or renewing — an AI interview assessment tool.