AI in Measurement Industry Practice: Perspectives from Assessment Organizations
November 10 @ 1:00 pm – 2:00 pm EST
This is presented by the Training & Development Committee
This panel asks practitioners who build and deploy measurement systems what has actually changed in their work because of AI, and what skills the field will need as a result. The discussion is about what these systems require of the organizations running and researching them, what they have made easier or harder, how things will change, and how NCME members can factor these things into their changing skill sets. The session is conversational rather than presentation-heavy. Each panelist offers brief opening remarks, followed by a moderated discussion built around prepared questions and questions from the audience. The broader goal of the series is to give NCME members opportunities between annual meetings to engage with timely issues relevant to their research, teaching, and professional practice.
About the Presenters
Ikkyu Choi is Research Director at ETS Research Institute, where he leads efforts to ensure AI is used responsibly in educational assessment. His work focuses on developing AI-driven solutions that are valid, reliable, and fair to help advance innovation without compromising integrity. Ikkyu received his Ph.D. in Applied Linguistics from the University of California, Los Angeles, and his research combines statistical modeling, machine learning, and natural language processing to develop new capabilities in educational measurement. He has published extensively across disciplines and was honored with the 2019 Best Article Award from the International Language Testing Association.
Matthew (Matt) S. Johnson is Managing Director, Innovations Research at the ETS Research Institute, where he is the Director of the Center for Responsible AI for Learning and Assessment. His work centers on fairness and validity methodology for AI scoring and test security, alongside human-in-the-loop approaches, multimodal data, and the integration of psychometric/statistical methods (e.g., IRT) with modern ML systems.
