Name
Beyond the Black Box: Validating Fairness and Trust in AI-Scored Oral Assessments
Description
AI-scored speaking assessments are scaling faster than the fairness and validity evidence behind them. This session offers a practical, vendor-neutral framework for evaluating AI oral scoring on three dimensions - validity, fairness, and transparency - drawn from real deployment experience in a production psychometric environment. Attendees leave with a validation checklist for auditing subgroup performance and score explainability, and a sharper standard for what to demand from vendors and internal teams before trusting AI scores in high-stakes decisions.
Speakers
Primary Topic
Trust, Ethics, and Policy in Assessment
Session Area
Certification & Licensure (C&L) Division, Technology-Based Assessment Committee, Equity Committee