Large Language Models for Data Visualization

This book offers the first comprehensive guide to understanding and applying large language models (LLMs) in data visualization. As LLMs continue to redefine how people interact with data, the book explores how they have made natural language a powerful interface for visualization, analysis, and reasoning. This book brings together the foundations of modern language models with the rapidly evolving research that is reshaping how visualizations are created, explored, communicated, and made accessible. Beginning with the principles behind transformers, multimodal LLMs, prompt engineering, and agentic AI, the book builds the knowledge needed to understand today’s AI systems before exploring their applications throughout the visualization pipeline. It covers visualization understanding and interaction, chart question answering, conversational and multimodal interfaces, visualization generation and editing, automated insight and story generation, human–LLM collaborative authoring, and AI-powered accessibility and inclusive visualization. Beyond applications, the book examines the capabilities, limitations, and trustworthiness of LLM-powered visualization systems, addressing topics such as hallucinations, reasoning errors, evaluation, responsible AI, human–AI collaboration, personalization, multilingual visualization, and future research opportunities. Richly illustrated with real-world examples, benchmarks, and case studies, this book serves as both an accessible introduction and an authoritative reference for researchers, graduate students, educators, and practitioners seeking to build the next generation of intelligent, trustworthy, and human-centered visualization systems.