AI for Official Statistics

This book exists to help move official statistics beyond the traditional, into applied machine learning, and onto GenAI.

Federal statistical agencies collect, process, and publish some of the most consequential data in the world. The methods behind that work are evolving. Machine learning, large language models, and agentic AI are not future possibilities; they are current tools with real applications in survey operations, data processing, and statistical production.

This book covers what those methods are, how they work, when they apply, and when they do not. Every chapter includes plain-language explanation, practical context for federal operations, and guidance on explaining methods to leadership and stakeholders.

All examples use public datasets from the Census Bureau, Bureau of Labor Statistics, and other federal sources. No proprietary data, no restricted access required. Full runnable code for every chapter is available in the companion repository: https://github.com/brockwebb/ai4stats