Free AIGP Exam Practice Questions

Sample Practice Questions

Try these 10 sample questions from the AIGP Exam practice bank. Each includes the answer and explanation. Continue with the interactive practice above to review the full question set.

1. 1Which feature is part of the NIST AI RMF description of an AI system?

   1. A. Generating predictions, recommendations, or decisions for given objectives
   2. B. Requiring a human to specify every output in advance
   3. C. Operating without any objectives or environmental influence
   4. D. Producing only physical actions rather than informational outputs

   Answer: A

   Explanation: "generate outputs such as predictions, recommendations, or decisions influencing real or virtual environments." The description includes informational outputs and varying autonomy; it does not require fully scripted outputs or exclusively physical actions.

   Reference: Artificial Intelligence Risk Management Framework (AI RMF 1.0) — Executive Summary, p. 1

2. 2Under the GDPR, how should a controller determine the personal data needed to train an AI system for a specified purpose?

   1. A. Collect all available data because larger datasets always justify broader processing
   2. B. Limit data to what is adequate, relevant, and necessary for that purpose
   3. C. Collect unrelated personal data if the model is statistically accurate
   4. D. Retain every possible feature unless an individual specifically objects

   Answer: B

   Explanation: "adequate, relevant and limited to what is necessary in relation to the purposes for which they are processed" Data minimisation requires a purpose-linked necessity assessment. Scale, predictive accuracy, or absence of objection is not a substitute.

   Reference: Regulation (EU) 2016/679 — General Data Protection Regulation — Article 5(1)(c)

3. 3Before building an AI system, a project team has chosen an algorithm but has not explained what organizational problem it solves. Which design activity is missing?

   1. A. Selecting a larger training batch
   2. B. Adding model parameters before evaluating the use case
   3. C. Defining the business value and context of use
   4. D. Increasing the number of benchmark datasets

   Answer: C

   Explanation: "The business value or context of business use has been clearly defined" A technical choice does not establish a business use case. More data, parameters, or benchmarks cannot supply the missing purpose.

   Reference: Artificial Intelligence Risk Management Framework (AI RMF 1.0) — MAP 1.4, p. 26

4. 4A business needs a model to produce new draft text rather than assign records to predefined classes. Which model capability directly matches the requirement?

   1. A. Generative AI
   2. B. Binary classification
   3. C. Clustering
   4. D. Numeric regression

   Answer: A

   Explanation: "Generative AI is a class of models that creates content from user input." Content generation is the requested function. Classification assigns categories, clustering groups patterns, and regression predicts numeric values.

   Reference: What is Machine Learning? — Generative AI

5. 5Why does AI governance need to consider social context as well as model design?

   1. A. Social context determines only the cost of model training
   2. B. Risks can arise through interactions between technical systems and human behavior
   3. C. A technically accurate model has the same impacts in every setting
   4. D. Social factors affect adoption but cannot affect AI risk

   Answer: B

   Explanation: "AI systems are inherently socio-technical in nature, meaning they are influenced by societal dynamics and human behavior." Technical performance alone does not determine effects. Human behavior and use conditions can change risks, rather than merely training cost or adoption.

   Reference: Artificial Intelligence Risk Management Framework (AI RMF 1.0) — Executive Summary, p. 1

6. 6A GDPR controller wants to reuse customer data for AI training that is incompatible with the original collection purpose. Which principle directly raises this issue?

   1. A. Storage limitation
   2. B. Integrity and confidentiality
   3. C. Purpose limitation
   4. D. Data portability

   Answer: C

   Explanation: "collected for specified, explicit and legitimate purposes and not further processed in a manner that is incompatible with those purposes" The incompatibility concerns purpose limitation. The other principles and right address retention, security, or transferability rather than compatibility of purposes.

   Reference: Regulation (EU) 2016/679 — General Data Protection Regulation — Article 5(1)(b)

7. 7Staff operate an AI service-allocation system, while applicants experience its decisions. Which proposed impact-assessment boundary recognizes harm to people who are not direct users?

   1. A. Limit the assessment to staff who operate the interface
   2. B. Limit the assessment to the supplier's model developers
   3. C. Limit the assessment to the team purchasing the system
   4. D. Include the applicants affected by allocation decisions

   Answer: D

   Explanation: "communities or other sub-groups who may be harmed are not always direct users of a system." Affected people can be harmed without directly using the interface; operator-only, developer-only and purchaser-only boundaries omit them.

   Reference: Artificial Intelligence Risk Management Framework (AI RMF 1.0) — Section 1.2.1, p. 6

8. 8A project has examples paired with known correct outcomes and wants to learn predictions from those pairs. Which learning approach fits?

   1. A. Grouping examples by similarity without known outcomes
   2. B. Learning predictions from examples with known correct outcomes
   3. C. Learning actions from rewards received through interaction
   4. D. Applying hand-written rules without learning from examples

   Answer: B

   Explanation: "data with the known correct results." Known correct outcomes provide the supervision signal. Unlabeled grouping, rewards from interaction and fixed rules use different mechanisms.

   Reference: What is Machine Learning? — Supervised learning

9. 9An AI service must perform as required without failure over a specified operating period under stated conditions. Which characteristic is being assessed?

   1. A. Transparency
   2. B. Interpretability
   3. C. Reliability
   4. D. Privacy

   Answer: C

   Explanation: "Reliability is a goal for overall correctness of AI system operation under the conditions of expected use and over a given period of time, including the entire lifetime of the system." The time interval and expected operating conditions identify reliability. Information availability, output meaning, and personal autonomy address different characteristics.

   Reference: Artificial Intelligence Risk Management Framework (AI RMF 1.0) — Section 3.1, p. 13

10. 10Which statement about an Article 6 GDPR lawful basis for AI processing is correct?

    1. A. Commercial benefit by itself always establishes a lawful basis
    2. B. Consent is the only possible basis for any AI processing
    3. C. Accurate model outputs remove the need for a lawful basis
    4. D. At least one applicable Article 6 basis must support the processing

    Answer: D

    Explanation: "Processing shall be lawful only if and to the extent that at least one of the following applies:" Article 6 lists multiple possible bases with their own conditions. AI accuracy and commercial benefit do not independently waive those conditions.

    Reference: Regulation (EU) 2016/679 — General Data Protection Regulation — Article 6(1)