Rethinking Assessment in the Age of AI: From Snapshots to Continuous Evidence Infrastructure

Advances in AI systems—including agentic workflows, persistent memory, and context-aware models—are transforming learning into a more continuous, adaptive, and multimodal process that unfolds across dialogue, simulation, collaboration, and real-world application over time. At the same time, higher education is moving toward more authentic, skill-based, and portfolio-driven approaches to assessment. Yet most assessment systems remain episodic and disconnected from how learning actually occurs, capturing isolated snapshots rather than learner development over time. This panel brings together researchers and practitioners to explore how assessment must evolve in response. Topics include process-based and multimodal assessment, longitudinal learner records, AI-supported feedback and evaluation, and emerging forms of AI-native assessment designed for continuous, context-aware learning environments. Panelists will also discuss the infrastructure required to support these approaches, including interoperability, learner-centered evidence systems, memory architectures, and embedded evaluation frameworks that allow evidence of learning to accumulate, persist, and remain interpretable across contexts over time. By connecting perspectives from AI, learning analytics, assessment, and educational practice, the session aims to advance a more integrated vision of assessment—one that reflects how learning actually happens and supports richer, more meaningful representations of learner growth.