In this episode of Tried and Tested, host Isabelle Gonthier speaks with Dr. Kristen DiCerbo, Chief Learning Officer at Khan Academy and one of Time’s top 100 people influencing the future of AI in 2024. Drawing on her background in school psychology, learning science, edtech, and assessment design, Kristen explores how AI is changing what assessment professionals can observe, interpret, and use as evidence of what learners know and can do.

As assessment programs look for more authentic ways to measure real-world skills, Kristen explains why better tasks are only part of the equation. The conversation covers task models versus evidence models, the importance of closing the inferential distance, lessons from simulation and game-based assessment, Khan Academy’s “Explain Your Thinking” work, and why new measurement approaches should often begin in lower-stakes formative environments before moving into higher-stakes use.

What you’ll learn:

• How AI can surface evidence of learner thinking beyond a single answer
• Why valid assessment starts with separating tasks from evidence
• What “closing the inferential distance” means for assessment design
• Where simulations, games, and open-ended tasks can create noise instead of signal
• Why new AI measurement approaches should start in low-stakes settings
• How to balance innovation with structure, validity, and defensibility
• What multimodal AI and on-device models could mean for future assessment

Who should listen:

• Assessment and credentialing leaders exploring AI
• Psychometricians and test developers
• Certification and licensure program owners
• Edtech and learning science teams
• Product leaders building digital assessment experiences
• Educators working with formative or performance-based assessment