What It Takes to Make AI Work

AI performance is only as strong as the data behind it. Weak data does not improve with better models. On this episode of Deconstructing Data, Jessie and David welcome Fern Halper, Founder of AI Foundations Group and VP of Research at TDWI, to discuss why data quality, structure, and governance determine whether AI delivers real outcomes. Fern works closely with executives and boards to help organizations navigate AI adoption with a focus on data, analytics, and strategy. She is the author of Data Makes the World Go ’Round, which explores the relationship between data, technology, and trust in successful AI initiatives. Her experience spans leadership roles at Bell Laboratories, AT&T, and Lucent Technologies, along with years of research into how companies implement data-driven systems at scale. In this conversation, the discussion focuses on the foundational role data plays in AI performance. Fern explains why many organizations struggle not because of AI itself, but because the underlying data is incomplete, unstructured, or unvalidated. The conversation explores how teams should approach newer data types, including unstructured signals, and why managing and validating this data is critical for accuracy and reliability. It also examines the role of governance, and how inconsistent or poorly managed data can create risk across AI-driven workflows. The episode closes with a discussion on how to think about the modern tech stack, and how to align infrastructure with data quality, validation, and performance outcomes. Topics include: 1. Building a Strong Data Foundation for AI 2. Managing and Validating Unstructured Data Signals 3. Governing Data for Accurate AI Performance 4. Tech Stack