AWS-AIF Exam Prep Free practice test →

Free AWS-AIF Practice Questions

10 free, exam-style AWS Certified AI Practitioner (AIF-C01) (AWS-AIF) practice questions with answers and explanations. No signup required. Work through them below, then take the full free AWS-AIF practice test to study every exam domain.

These 10 free AWS-AIF questions are organized by exam domain, so you can see how each part of the AWS Certified AI Practitioner (AIF-C01) blueprint is tested. Reveal the answer and explanation under each question.

Domain 1: Fundamentals of AI and ML

Question 1

A retail company wants to group their customers into segments based on purchasing behavior, but they have no predefined categories. Which ML approach is MOST appropriate?

  1. Supervised classification
  2. Supervised regression
  3. Unsupervised clustering
  4. Reinforcement learning
Show answer & explanation

Correct answer: C - Unsupervised clustering

Question 2

A company wants to predict whether a customer will churn (yes/no) based on their usage patterns. Which ML technique is MOST appropriate?

  1. Regression
  2. Classification
  3. Clustering
  4. Dimensionality reduction
Show answer & explanation

Correct answer: B - Classification

Domain 2: Fundamentals of Generative AI

Question 3

A company uses a foundation model and notices it sometimes produces outputs that contain biased stereotypes. What is the MOST likely root cause?

  1. The model's inference hardware is insufficient
  2. The model learned and reflects biases present in its pre-training data
  3. The temperature setting is too low
  4. The context window is too small
Show answer & explanation

Correct answer: B - The model learned and reflects biases present in its pre-training data

Question 4

A user asks a GenAI model about events that occurred last week, but the model provides outdated information. What is the MOST likely cause?

  1. The model's temperature is too high
  2. The model's knowledge is limited to its training data cutoff date
  3. The model is hallucinating
  4. The network connection is slow
Show answer & explanation

Correct answer: B - The model's knowledge is limited to its training data cutoff date

Domain 3: Applications of Foundation Models

Question 5

A company wants their foundation model to respond in a specific JSON format for their API integration. Which customization approach is MOST appropriate and cost-effective?

  1. Fine-tuning the model on JSON examples
  2. Continued pre-training on JSON data
  3. Prompt engineering with clear format instructions
  4. RAG with a JSON document database
Show answer & explanation

Correct answer: C - Prompt engineering with clear format instructions

Question 6

A company wants to reduce hallucinations when the FM answers questions about their products. Which customization approach is MOST effective?

  1. Increasing the temperature parameter
  2. Using RAG to ground responses in actual product documentation
  3. Making the model larger
  4. Using continued pre-training only
Show answer & explanation

Correct answer: B - Using RAG to ground responses in actual product documentation

Domain 4: Guidelines for Responsible AI

Question 7

A company fine-tunes a model and the resulting model has MORE bias than the original base model. What is the MOST likely cause?

  1. The fine-tuning process always increases bias
  2. The fine-tuning data contained biases that were introduced into or amplified by the model
  3. The model became too large
  4. The temperature was too high
Show answer & explanation

Correct answer: B - The fine-tuning data contained biases that were introduced into or amplified by the model

Question 8

A company's GenAI model passes all bias tests during development but shows biased behavior after 6 months in production. What MOST likely happened?

  1. The tests were wrong
  2. Real-world data patterns have shifted (data drift), causing the model to exhibit bias it didn't show during testing
  3. The model improved on its own
  4. AWS changed the model
Show answer & explanation

Correct answer: B - Real-world data patterns have shifted (data drift), causing the model to exhibit bias it didn't show during testing

Domain 5: Security, Compliance, and Governance for AI Solutions

Question 9

A company needs to maintain an audit trail of all data pipeline operations including data collection, transformation, and model training steps. Which is the BEST approach?

  1. Manual record keeping in spreadsheets
  2. Automated logging with CloudTrail, SageMaker Pipelines tracking, and Glue Data Catalog
  3. No documentation needed
  4. Only track the final model output
Show answer & explanation

Correct answer: B - Automated logging with CloudTrail, SageMaker Pipelines tracking, and Glue Data Catalog

Question 10

A healthcare company building an AI diagnostic tool needs to ensure end-to-end governance. Which combination is MOST comprehensive?

  1. Only model accuracy testing
  2. HIPAA compliance + data encryption + access controls + bias testing + human oversight + audit trails + model documentation
  3. Only cost optimization
  4. Only model selection
Show answer & explanation

Correct answer: B - HIPAA compliance + data encryption + access controls + bias testing + human oversight + audit trails + model documentation

Ready for the real thing?

Practice hundreds more AWS-AIF questions with instant scoring, weak-area drills, and full exam simulations.

Start the free practice test See pricing