From Isolated to Unified: Building Trustworthy AI on Decision-Grade Institutional Data

Artificial intelligence is only as trustworthy as the data that powers it. As institutions accelerate AI adoption, many are discovering that traditional data governance alone is insufficient to support responsible and reliable AI deployment. This session explores how Carnegie Mellon University integrated data governance and AI governance into a unified operational framework designed to enable trustworthy AI. We will discuss how aligning governance structures across data stewardship, analytics, and AI initiatives creates the foundation for reliable natural language interfaces, reduced hallucination risk, and scalable institutional AI innovation. Drawing on higher education practice and emerging governance guidance, the presenters will introduce a practical framework for transforming distributed institutional data into AI-ready data products. Participants will learn how establishing data maturity standards, audited data pipelines, and SQL-augmented AI architectures enables accurate natural language interrogation of institutional data. Attendees will leave with a higher ed-tailored blueprint for moving from siloed data management to unified data and AI governance, enabling trustworthy analytics, natural language access to institutional data, and safer deployment of AI workflows and agents.