What works?! A National R&D Infrastructure for Trustworthy AI

Across campuses and classrooms, ambient AI systems are capturing context continuously - conversations, workflows, assessments, and interactions - building persistent learner memory and adapting in real time. Learning is shifting from discrete sessions to continuous streams, from static snapshots to longitudinal memory, from isolated tasks to agentic collaboration, and from standalone platforms to interconnected ecosystems.

But ambient capture changes everything: from how data flows, to how assessment works, to what learner autonomy really means.

As AI becomes embedded in the air of campus life, how do we design systems that empower rather than surveil? Scale personalization without eroding agency? Build products that are interoperable across a continuous intelligence layer?