- What This Cheat Sheet Covers
- The Exam at a Glance
- The Four Domains in One Pass
- Prompt Patterns Worth Memorizing
- Ethics, Law, and Societal Impact Quick Hits
- Scoring and Question Format Facts
- Vouchers and Booking Mechanics
- Eligibility and Test-Day Rules
- Validity and Renewal
- Who Values This Credential
- Sequencing Your Review
- Frequently Asked Questions
- Generative AI Foundations is a Pearson exam delivered through Certiport, with a 50-minute timer and a passing score of 700 on a 1-1000 scale.
- Four domains: Generative AI Methods and Methodologies, Basic Prompt Engineering, Prompt Refinement, and Ethics, Law, and Societal Impact.
- Certiport's single-exam voucher is USD 72 (USD 83 with a retake) and works only at Certiport Authorized Testing Centers.
- The question count is documented inconsistently: 40-45 per the Certiport tutorial, 45-50 per the Mindhub page. Expect variation by exam form.
What This Cheat Sheet Covers
This page condenses what you need to know about the Generative AI Foundations exam into a single scannable review. "GAF" here means Pearson's Generative AI Foundations certification, developed and administered through Certiport as part of its Critical Career Skills (CCS) family. Other credentials share the same three letters, so double-check that any outside material you use refers to this exam and not a different one.
Use the sheet in two ways: as a first-pass orientation before you start studying, and as a final-week refresher once you have worked through the full material. For depth on any single topic, the GAF study guide and the complete guide to all four content areas go further than a one-pager can. If you want to test recall right away, the GAF practice test is the fastest way to find your weak spots.
The Exam at a Glance
| Item | What to Know |
|---|---|
| Issuer / administrator | Pearson; developed and delivered through Certiport (a Pearson VUE business) under Critical Career Skills |
| Timed duration | 50 minutes, excluding the tutorial and administrative seat time |
| Passing score | 700 on Certiport's 1-1000 scaled-score system |
| Question count | 40-45 per the Certiport tutorial; 45-50 per the Mindhub voucher page; totals differ by exam form |
| Format | Proctored, computer-based knowledge assessment with answer selection, flagging, and review before final submission |
| Orientation | Brand-agnostic; described as best suited for ages 14 and older |
| Domains | Four, with no published percentage weights |
| Validity | Five years for certifications earned on or after June 18, 2025 |
Two details trip people up. First, the 50 minutes covers the exam itself; the tutorial screens and check-in time sit outside that clock. Second, the question count is genuinely inconsistent across official sources, so do not anchor your pacing plan to a single number. Plan for the higher end of the range and leave time to review flagged items. Our difficulty guide explains how pacing interacts with the way questions are worded.
The Four Domains in One Pass
The domain names below reproduce the headings in Certiport's Generative AI Foundations exam objectives. Because no weights are published, treat all four as fair game and do not skip the one that feels least technical.
Domain 1: Generative AI Methods and Methodologies
The conceptual foundation. Expect questions about what generative AI is, how it differs from other forms of AI, and how generative tools produce text, images, and other content.
- Distinguish generative AI from traditional rule-based software and from predictive systems
- Recognize common generative output types and the kinds of tasks each suits
- Understand why outputs can be fluent yet wrong, and what that means for verification
- Know the role of training data and how it shapes model behavior
Domain 2: Basic Prompt Engineering
The practical skill of writing instructions that get usable results from a generative tool.
- Clear, specific instructions outperform vague requests
- Supplying context, audience, and desired format improves relevance
- Assigning a role or perspective can steer tone and depth
- Examples inside a prompt help the tool match a pattern
Domain 3: Prompt Refinement
What you do after the first response disappoints. This domain rewards an iterative mindset.
- Diagnose why an output missed the mark: missing context, ambiguity, or conflicting constraints
- Adjust one element at a time so you can see what changed the result
- Use follow-up prompts to narrow, expand, reformat, or correct a response
- Evaluate outputs critically instead of accepting the first draft
Domain 4: Ethics, Law, and Societal Impact
Responsible use, and the area where many candidates underestimate how much reading comprehension matters.
- Bias, fairness, and representation in generated content
- Privacy and the risk of entering sensitive information into a tool
- Intellectual property and attribution questions around AI-generated work
- Transparency, accountability, and the broader effect on work and society
Prompt Patterns Worth Memorizing
Because the exam is brand-agnostic, it tests principles rather than the buttons of any one product. Two domains out of four center on prompting, so it pays to be fluent in the vocabulary and able to judge which of several prompts is strongest.
The Elements of a Strong Prompt
- Task: the action you want performed, stated plainly
- Context: background the tool cannot guess
- Audience and tone: who the output is for and how it should sound
- Format: list, table, paragraph, length limit, or structure
- Constraints: what to include, avoid, or prioritize
How Refinement Questions Tend to Work
Scenario-style items often show an initial prompt, a disappointing output, and ask what to change. The best answer usually addresses the specific gap, such as adding missing context or tightening a vague instruction, rather than rewriting everything or abandoning the tool. When two options both seem reasonable, prefer the one that is targeted and testable.
Ethics, Law, and Societal Impact Quick Hits
This domain tests judgment. Questions tend to describe a situation and ask what a responsible user should do, so read the scenario carefully for the risk it is hiding.
- Verify before you rely. Generative tools can state falsehoods confidently. Check facts against trustworthy sources, especially for anything consequential.
- Protect sensitive data. Personal, confidential, or proprietary information should not be pasted into a tool without understanding how that input is handled.
- Watch for bias. Outputs can reflect patterns in training data. Review content for stereotypes and uneven representation.
- Respect ownership. Questions about copyright, attribution, and disclosure reward cautious, transparent answers.
- Stay accountable. The person who publishes or acts on AI-generated content remains responsible for it.
Key Takeaway
On ethics items, the correct answer is rarely the most permissive one. Look for the option that combines human oversight, verification, and transparency about AI involvement.
Scoring and Question Format Facts
The passing score is 700 on a scale of 1 to 1000. That is a scaled score, not a percentage of questions answered correctly, and Certiport notes that forms differ. So there is no clean conversion from "700" to "70 percent right," and you should not try to back into a target number of correct answers. Our passing score breakdown covers what scaled scoring does and does not tell you.
Inside the exam you can select answers, flag questions you are unsure about, and review before final submission. Use that review feature deliberately: answer every item on a first pass, flag the uncertain ones, and return to them with whatever time remains. Leaving items blank is rarely better than making your best reasoned choice.
If you are curious about how candidates actually perform, see what the available data shows on pass rates. Be wary of any source that quotes a precise figure without citing the issuer.
Vouchers and Booking Mechanics
This is the area where candidates most often lose money or time, because there are two voucher routes with different rules. Confirm which you are buying before you pay.
| Route | U.S. Price | Delivery Rule |
|---|---|---|
| Certiport single-exam voucher | USD 72 | Valid only through a Certiport Authorized Testing Center (in person or remotely) |
| Certiport voucher with retake | USD 83 | Same CATC-only restriction |
| Pearson Mindhub voucher | USD 85 | Country restrictions apply; must match the country chosen at checkout |
| Mindhub Pro listing | USD 85 (web and member price) | Same country-matching requirement |
Prices above are U.S. figures and exclude any proctoring charge a testing center may add. For a fuller look at what to budget, including optional practice materials, read the GAF certification cost breakdown. For scheduling logistics and deadlines, see GAF exam dates and scheduling.
Eligibility and Test-Day Rules
There is no bachelor's degree requirement. The issuer instead expects roughly 150 combined hours of instruction and hands-on use of generative AI tools, along with familiarity with productivity applications. Those hours describe preparation, not the exam timer and not a mandatory course length. The full picture is in our requirements and eligibility guide.
Test-day rules follow Certiport Authorized Testing Center policy:
- The exam is proctored
- You need valid identification
- Candidates under 18 need parental consent
- Books and personal devices must remain inaccessible during the exam
That last point matters for a cheat sheet: it is a study aid, not something you can bring to the exam. Treat it as a memory prompt and review it beforehand.
Validity and Renewal
CCS certifications earned on or after June 18, 2025 are valid for five years. Renewal means purchasing a new voucher and passing the current version of the exam. Certifications earned earlier retain their existing status. If you plan to list the credential on a resume or profile, note the date you earned it so you know when the five-year window closes.
Who Values This Credential
Generative AI Foundations is an entry-level, brand-agnostic credential, so it signals baseline literacy rather than specialized engineering skill. It tends to be most relevant for students, career changers, and professionals in office, administrative, marketing, education, and operations roles who now use AI tools in daily work and want a verifiable way to show they can do so competently and responsibly. Employers and schools that want evidence of AI fluency without requiring a technical background are the natural audience.
Be realistic about what the credential does on its own. It supports a resume and a conversation; it does not guarantee a role or a particular pay level. For a grounded look at earning potential and the hiring picture, see the salary analysis and the ROI analysis.
Sequencing Your Review
Since the objectives carry no published weights, a sensible plan gives each domain real attention and front-loads the one that builds the vocabulary for the rest. Here is one way to order a four-week review, assuming you have already been using generative tools hands-on.
Generative AI Methods and Methodologies
- Lock in core terminology; every other domain builds on it
- Practice explaining the difference between generative and traditional AI in your own words
Basic Prompt Engineering
- Write and compare prompts for the same task at different levels of specificity
- Memorize the elements of a strong prompt
Prompt Refinement
- Take weak outputs and fix them one adjustment at a time
- Practice diagnosing why a response missed
Ethics, Law, and Societal Impact, then full practice
- Work scenario questions on bias, privacy, and ownership
- Finish with timed practice sets and review every miss
Prompting skills are easiest to retain when you do them, so keep a generative tool open while you study rather than only reading. Pair this schedule with the broader plan in the GAF study guide, and when you want realistic timed repetition, run sets on the practice test site.
Frequently Asked Questions
Official sources disagree. Certiport's tutorial specifies 40-45 questions, while the Mindhub voucher page estimates 45-50. Certiport also states that totals differ by exam form, so prepare for the higher range and manage your 50 minutes accordingly.
You need a scaled score of 700 on Certiport's 1-1000 system. It is not a percentage of correct answers, so do not assume 700 equals 70 percent.
No. Certiport vouchers are valid only for delivery through a Certiport Authorized Testing Center, in person or remotely, and cannot be redeemed through OnVUE or a Pearson VUE Testing Center. Verify compatibility with your chosen channel before purchasing.
Certifications earned on or after June 18, 2025 are valid for five years. Renewal requires a new voucher and passing the current exam. Earlier nonexpiring certifications keep their existing status.
No bachelor's degree is required. The issuer expects about 150 combined hours of instruction and hands-on use of generative AI tools, plus familiarity with productivity applications. That figure describes recommended preparation, not a mandatory course or the exam length.