AI Interviewing and Credibility Assessment

AI interviews are increasingly popular tools for use in employee selection and security vetting. Traditional interview methods are inconsistent: interviewers vary in skill, responses vary in quality, and human judgement of honesty is unreliable, often performing little better than chance. At the same time, AI-powered interviewing tools are entering the market without a strong independent evidence base to guide their use.

This project aims to fill that gap by rigorously evaluating AI interviewing technology. It seeks to evaluate AI interview delivery methods, identify and evaluate signals for AI-based assessment, critically evaluate signals of deception, and contribute to the ethical evaluation of their use.

Funded by the Office of National Intelligence.