GAF Exam Domains 2026: Complete Guide to All 4 Content Areas

How the Official Outline Is Built

The Generative AI Foundations (GAF) exam is a Pearson credential developed and administered through Certiport, a Pearson VUE business, as part of the Critical Career Skills (CCS) family. Its content is defined by Certiport's published Generative AI Foundations Exam Objectives document, which lists four domains:

1. Generative AI Methods and Methodologies
2. Basic Prompt Engineering
3. Prompt Refinement
4. Ethics, Law, and Societal Impact

One detail that surprises many candidates: the outline does not publish percentage weights. Many certifications tell you that a given domain is, say, a fixed share of scored items. This one does not, so any site claiming precise domain percentages for this exam is guessing. The practical consequence is simple. You should prepare all four domains to a competent level rather than gambling on a heavy domain and skipping a light one.

The recommended preparation profile is also worth knowing. Certiport expects roughly 150 combined hours of instruction and hands-on use of generative AI tools, along with familiarity with productivity applications. That figure describes the background a well-prepared candidate typically has, not the length of the exam and not a mandatory course. No bachelor's degree is required, and the exam is described as best suited for ages 14 and older. The full eligibility picture is covered in GAF Requirements 2026: Eligibility, Prerequisites & How to Qualify.

Domain 1: Generative AI Methods and Methodologies

This is the conceptual foundation of the exam. Think of it as the "what is happening under the hood and why does it matter to a user" domain. You are not expected to build models or write training code. You are expected to speak the vocabulary and reason about how generative systems behave.

What a question here tends to look like

Expect scenario-flavored multiple-choice items rather than math. A typical question might describe a business or classroom situation and ask which type of tool or approach fits, or ask you to identify why a model produced a confident but inaccurate statement. The correct answer usually hinges on understanding that these systems predict plausible content rather than retrieve verified facts.

Concepts worth mastering

• Probabilistic generation: Know why identical prompts can yield different answers and what that implies for reliability.
• Training versus using: Understand the distinction between how a model was built and how you interact with it as an end user.
• Fit for purpose: Be ready to say when generative AI is a reasonable choice and when a conventional tool, search, or human review is better.
• Limitations: Hallucination, outdated knowledge, and sensitivity to phrasing are recurring themes that feed directly into Domains 3 and 4.

If you want a one-page memory aid for the vocabulary in this domain, the GAF Cheat Sheet 2026: One-Page Review of Must-Know Facts is a good companion.

Domain 2: Basic Prompt Engineering

Domain 2 is where the exam becomes hands-on. A prompt is the instruction you give a generative model, and this domain tests whether you can write one that is clear enough to get a useful first result.

Weak prompt versus strong prompt

The exam favors recognition and judgment. You may be shown two or more prompts for the same goal and asked which is most likely to produce a useful result, or asked which element is missing from an underperforming prompt.

Beyond the single instruction

Basic prompt engineering also covers the idea that a prompt can include constraints (what to avoid), a desired reasoning approach, and reference material the model should use. At the foundations level, you are expected to recognize these tools and know why they help, not to master advanced prompting frameworks.

Domain 3: Prompt Refinement

If Domain 2 is writing a good first prompt, Domain 3 is what you do when the first answer is not good enough. This domain treats prompting as an iterative conversation rather than a one-shot command.

The diagnosis skill

The most testable idea in this domain is diagnosis. A question may present a prompt, the response it produced, and a complaint such as "the answer was too long and off-topic," then ask for the best next step. The strongest answers tie the symptom to a specific fix: add a length limit for verbosity, add audience context for tone mismatch, add a format instruction for disorganized output.

Refinement is not the same as rewriting from scratch

A common trap is choosing an answer that discards the whole prompt when a targeted adjustment would do. Foundations-level refinement favors incremental changes: keep what worked, change what did not. Questions may also probe whether you recognize that some failures, such as invented facts, cannot be prompted away entirely and require human verification.

Domain 4: Ethics, Law, and Societal Impact

The fourth domain moves from technique to responsibility. It is easy to underestimate because it contains little hands-on skill, yet it is where a candidate with strong prompting instincts can lose points by relying on common sense instead of the specific concepts the exam expects.

Scenario reasoning over memorization

Because laws and policies vary by place and change over time, expect the exam to test principles rather than jurisdiction-specific statutes. A question might describe an employee pasting confidential client data into a public tool, or a student submitting generated text without disclosure, and ask which concern is most relevant. Identify the underlying principle (privacy, attribution, accuracy, fairness) and the answer usually follows.

Exam Format, Scoring, and Delivery

Knowing the domains is only half the picture. You also need to understand the testing mechanics so nothing about the format surprises you.

Reading the score correctly

A 700 is a scaled score, not 70 percent. Scaling accounts for differences between exam forms, so you cannot convert 700 into a fixed number of correct answers. The practical lesson is to aim for solid mastery of every domain instead of trying to calculate a minimum. A deeper explanation is in GAF Passing Score 2026: Exactly What You Need to Pass.

Vouchers and delivery channels

Pricing and delivery are easy to get wrong, so read them carefully:

• In the U.S., Certiport lists a single-exam voucher at USD 72, or USD 83 with a retake, plus any proctoring charge at the testing center.
• Those Certiport vouchers are valid only for Certiport Authorized Testing Center (CATC) delivery, in person or remotely. They cannot be redeemed through OnVUE or a Pearson VUE Testing Center.
• A separate Pearson Mindhub voucher is listed at USD 85 and carries country restrictions that must match the country selected at checkout.
• Pearson's launch announcement mentions an OnVUE delivery route, but a current GAF-specific OnVUE booking process could not be verified, so confirm voucher compatibility with your chosen channel before paying.

CATC rules require proctoring, valid identification, and parental consent for candidates under 18, and books and personal devices must stay inaccessible during the exam. For the full cost picture, read GAF Certification Cost 2026: Complete Pricing Breakdown, and for scheduling logistics see GAF Exam Dates 2026: Testing Windows, Deadlines & Scheduling.

Sequencing the Four Domains in Your Prep

Because the outline is unweighted, a balanced plan beats a lopsided one. The order below follows how the domains build on each other: concepts first, then technique, then iteration, then responsibility. Adjust the timeline to your own background and available hours.

Domain 4 sits last on purpose: it reuses everything you saw earlier, since bias, hallucination, and verification are limitations you will already have encountered. For a fuller preparation framework, see the GAF Study Guide 2026: How to Pass on Your First Attempt, and for a realistic sense of effort, How Hard Is the GAF Exam? Complete Difficulty Guide 2026. Then pressure-test yourself with timed items on the GAF Exam Prep practice site.

Where These Skills Show Up at Work

The four domains map to everyday workplace behavior more than to a single job title. Employers rarely hire for "generative AI foundations" in isolation; they look for people who can use these tools productively and safely inside ordinary roles.

• Office and administrative roles: Drafting, summarizing, and reformatting content using prompt techniques from Domains 2 and 3.
• Marketing and communications: Producing first drafts and variations while applying the disclosure and accuracy habits from Domain 4.
• Education and training: Building materials and teaching responsible use, drawing on all four domains.
• Students and early-career candidates: Showing a verified baseline of AI literacy on a resume, since the exam suits learners aged 14 and older.

A credential demonstrates baseline competence rather than guaranteeing a particular outcome, so treat it as one signal among several. For earnings context see GAF Salary Guide 2026: Complete Earnings Analysis, and for the broader orientation What Is GAF?

Frequently Asked Questions

Mastering these four domains in order, concepts, technique, iteration, and responsibility, gives you a durable foundation rather than a set of memorized tricks. Revisit the GAF certification overview if you want to confirm how the credential fits your goals before you book.