How to use this book

Who this book is for

This is the study book for AI Literacy Foundation (ALF), AIDAN's first track, White Belt. It is written for anyone who uses, or plans to use, generative AI at work, whatever their role. Its primary readers are people who use AI as a tool: marketers, accountants, customer support advisers, teachers and public servants. Its secondary audience is managers, PMs and executives evaluating AI adoption. No coding experience is required. It does not cover writing code, system design or model training. Even when it cites developer documentation, it draws only on concepts that non-developers need to understand.

After reading the whole book and completing the exercises, you should be able to do five things. These are also the exam's objectives.

1. Explain where AI goes wrong in terms of tokens, the context window, probabilistic generation and the knowledge cutoff.
2. Choose the product form that fits the task: chat, projects, custom assistants, connectors, agents or API.
3. Improve output quality with clear instructions, context, examples and output formats, and revise instructions after reviewing the results.
4. Decide whether sensitive information may be entered, and verify AI-generated output against evidence.
5. Compare the value and risks of business use cases, and choose metrics that demonstrate their impact in numbers.

The book uses the four competencies of the AI Fluency Framework as its organizing foundation. Professors Rick Dakan and Joseph Feller developed the framework, and Anthropic worked with them to create a course based on it. The framework holds that four competencies must work together for interactions with AI to be effective, efficient, ethical and safe.

How this book is organized

The book consists of this guide, five chapters and four appendices. The chapters follow the exam domains in order, and each is written to be understandable on its own. Every chapter follows the same ten sections in the same order.

Appendix A is a glossary; Appendix B provides self-check answers and explanations; Appendix C is an exam day guide; and Appendix D lists the official documents cited in the book. If a term is unfamiliar when it first appears, look it up in Appendix A.

Four kinds of boxes appear in the text. Notes provide additional information, tips highlight principles to remember, warnings describe common incidents and misunderstandings, and examples contain exercises to try yourself.

The exam at a glance

The exam is weighted as follows. It is a good idea to spend more study time on Chapters 1 and 2, which carry greater weight.

Questions favor lasting principles over the locations of buttons in particular products. Even questions about a product ask, "What properties of this feature make it suitable for this situation?" This book therefore explains properties before product names.

What the questions look like

Questions come in five formats. Each format is scored differently, so get to know them in advance.

Most questions start with a short workplace scene. The scene includes an organization, a situation and one or two conditions in bold. The correct answer almost always meets both conditions together. An option that meets only one condition is a distractor designed to sound plausible. The following example was written separately for this book to illustrate the format; it is not an actual exam question.

[Example] The owner of a neighborhood florist, Wildflower Shop, wants AI to improve a customer notice.
The notice will be **handwritten on the chalkboard at the shop entrance** and **must not exceed three lines**.
Which request is the most appropriate?

In this example, "handwritten" means that formatting such as tables and emoji is not useful, while "three lines" is a length constraint. The correct request includes both conditions. Develop the habit of identifying the bold conditions first and interpreting what each rules out. Appendix C explains this technique in more detail.

Six-week study plan

This plan assumes 30–40 minutes of study a day, four times a week. Ideally, try the exercises with the tools you use at work, but use public or invented materials instead of company data.

Notation and sources

The book follows these principles.

• Verification dates: Product features, plans, data handling, usage policies and legislation were checked on each company's or institution's official pages on October 20–21, 2026. The pages are collected in each chapter's "Check the official docs" section and Appendix D.
• Principles first: AI products frequently change their names and features. Exact prices, model version numbers and plan-specific limits are therefore omitted unless essential. For the same reason, product descriptions state how a company describes something, while decision criteria are expressed as principles independent of products.
• Original names: Product and technology names such as ChatGPT, Claude, Gemini and API retain their original spelling. Terms are explained where they first appear.
• Invented organizations: Every company, person and number in workplace scenes and exercises is invented. Any overlap with a real company's name is coincidental.
• Legal content: Korea's Personal Information Protection Act (PIPA), AI Basic Act and Copyright Act were checked against the current provisions on the National Law Information Center portal. This book does not provide legal advice. Consult your organization's legal or privacy officer when an actual decision is needed.