Reading Guide

This primer provides a comprehensive introduction to synthetic data, covering everything from basic concepts to advanced applications. Each chapter is designed to be as standalone as possible, so you can jump to topics that interest you most. This guide helps you navigate the content based on your background and interests.

Quick Start Paths

Reader Type	Primary Focus	Recommended Reading Path
Beginners	Understanding fundamentals	What is Synthetic Data → Why Use Synthetic Data → Generation Methods → Quality Evaluation → Real World Deployments
Practitioners	Implementation & development	Why Use Synthetic Data → Generation Methods → Quality Evaluation → Privacy-preserving Synthesis → Practical Data Synthesis → LLM-driven Data Synthesis → Real World Deployments → Challenges and Risks
Decision Makers	Business value & strategy	Why Use Synthetic Data → Real World Deployments → Applications in Public Sector → Challenges and Risks → Outlook & Trends
Privacy Officers	Compliance & risk management	Privacy-preserving Synthesis → Real World Deployments → Applications in Public Sector → Challenges and Risks
Researchers	Latest developments & methods	LLM-driven Data Synthesis → Challenges and Risks → Outlook & Trends

Key Concepts & Navigation Elements

Essential Terms

These fundamental concepts appear throughout the primer. For detailed definitions, see our Glossary of Terms.

• Synthetic Data - Artificial data generated by algorithms that imitates the patterns of real data, such as survey responses, financial transactions, or sensor readings collected from actual people, events, or systems
• Synthesizer - The AI system that creates synthetic data by learning from real examples
• Generative Models - AI that can create new content (like text, images, or tabular data) after learning from examples
• Training Dataset - The real data used to teach the AI what patterns to copy
• Held-out Dataset - Real data kept separate to test how well the synthetic data works
• Fine-tuning - Customizing an AI model for a specific task or industry
• Prompt Engineering - Writing instructions to get AI to produce what you want
• Data Modality - The format of data (spreadsheets, photos, text, audio recordings, etc.)
• Privacy Preservation - Managing the risk of exposing people’s identifiable or sensitive information

Callout Boxes Guide

Throughout the primer, you’ll encounter four types of highlighted information:

Info Boxes

Provide additional context and clarifications that enhance your understanding of the topic. These contain supplementary details that are valuable but not essential to the main flow.

Tip Boxes

Offer practical tools, quick-start advice, and actionable recommendations. These help you get started with implementation or provide shortcuts to common tasks.

Example Boxes

Present concrete examples, case studies, and real-world applications. These illustrate concepts in practice and show how theoretical knowledge applies to actual scenarios.

Warning Boxes

Highlight important cautions, potential risks, and critical considerations. These alert you to common pitfalls, security concerns, or situations that require careful attention.

Getting Help

• Glossary of Terms - Definitions of technical concepts and terminology
• References - Cited research papers and additional resources
• Credits - Contributors, acknowledgments, and version history