Try 30 free AWS AI Practitioner practice questions
Real exam-style questions taken word for word from the AWS AI Practitioner study guide. Ten are below, with the full answer explanation. Enter your email and the full 30-question PDF is yours.
Unlock the full sampler
Ten questions to try right now
Work each one before you open the answer. These are the same style and depth as the AWS Certified AI Practitioner (AIF-C01) exam expects.
Question 1. A company wants to use large language models (LLMs) with Amazon Bedrock to develop a chat interface for the company's product manuals. The manuals are stored as PDF files. Which solution meets these requirements MOST cost-effectively?
- A. Use prompt engineering to add one PDF file as context to the user prompt when the prompt is submitted to Amazon Bedrock.
- B. Use prompt engineering to add all the PDF files as context to the user prompt when the prompt is submitted to Amazon Bedrock.
- C. Use all the PDF documents to fine-tune a model with Amazon Bedrock. Use the fine-tuned model to process user prompts.
- D. Upload PDF documents to an Amazon Bedrock knowledge base. Use the knowledge base to provide context when users submit prompts to Amazon Bedrock.
Show the answer
Answer: D. A knowledge base retrieves only the most relevant chunks for each question, which keeps token usage and cost low while covering the full manual set. Stuffing one PDF limits coverage, stuffing all PDFs inflates token cost per call, and fine-tuning is expensive and unnecessary for retrieving reference content.
Question 2. A company has a generative AI application that uses a pre-trained foundation model (FM) on Amazon Bedrock. The company wants the FM to include more context by using company information. Which solution meets these requirements MOST cost-effectively?
- A. Use Amazon Bedrock Knowledge Bases.
- B. Choose a different FM on Amazon Bedrock.
- C. Use Amazon Bedrock Agents.
- D. Deploy a custom model on Amazon Bedrock.
Show the answer
Answer: A. Amazon Bedrock Knowledge Bases add company context through retrieval at inference time without any training, which is the lowest-cost option. Switching models does not introduce company data, agents orchestrate multi-step actions rather than simply adding context, and deploying a custom model is far more expensive.
Question 3. A company uses an Amazon Bedrock large language model (LLM) in an application. During testing, the company observes different outputs from the same input. What is the MOST likely cause of this issue?
- A. The LLM is acting in a nondeterministic way.
- B. The guardrails of the LLM are not configured properly.
- C. The LLM has security vulnerabilities.
- D. The LLM is acting in a deterministic way.
Show the answer
Answer: A. Producing different outputs for the same input is the hallmark of nondeterministic behavior, which comes from sampling-based generation. Deterministic behavior would produce identical output every time, so D is the opposite of what was observed. Misconfigured guardrails and security vulnerabilities would not specifically cause varied responses to the same prompt, so A is the most likely cause.
Question 4. A company is using the Generative AI Security Scoping Matrix to assess security responsibilities for its solutions. The company has identified four different solution scopes based on the matrix. Which solution scope gives the company the MOST ownership of security responsibilities?
- A. Using a third-party enterprise application that has embedded generative AI features.
- B. Building an application by using an existing third-party generative AI foundation model (FM).
- C. Refining an existing third-party generative AI foundation model (FM) by fine-tuning the model by using data specific to the business.
- D. Building and training a generative AI model from scratch by using specific data that a customer owns.
Show the answer
Answer: D. The Generative AI Security Scoping Matrix assigns the most customer responsibility to the scope where the organization builds and trains its own model, because the company controls the data, the training pipeline, and the entire model lifecycle. Consuming a third-party app or using a third-party FM shifts most of the burden to the provider. Fine-tuning in option C adds some responsibility but still relies on a vendor base model, so it carries less ownership than building from scratch.
Question 5. A company has created a custom model by fine-tuning an existing large language model (LLM) from Amazon Bedrock. The company wants to deploy the model to production and use the model to handle a steady rate of requests each minute. Which solution meets these requirements MOST cost-effectively?
- A. Deploy the model by using an Amazon EC2 compute optimized instance.
- B. Use the model with on-demand throughput on Amazon Bedrock.
- C. Store the model in Amazon S3 and host the model by using AWS Lambda.
- D. Purchase Provisioned Throughput for the model on Amazon Bedrock.
Show the answer
Answer: D. Provisioned Throughput reserves dedicated capacity on Amazon Bedrock, and a steady, predictable per-minute request rate is exactly the workload where reserved capacity is most cost-effective. On-demand pricing works well for variable or burst traffic but becomes more expensive than provisioned capacity at sustained, consistent load. Self-hosting on EC2 or Lambda adds the operational overhead that Bedrock is designed to eliminate.
Question 6. A company wants to implement a large language model (LLM) based chatbot to provide customer service agents with real-time contextual responses to customers' inquiries. The company will use the company's policies as the knowledge base. Which solution will meet these requirements MOST cost-effectively?
- A. Retrain the LLM on the company policy data.
- B. Fine-tune the LLM on the company policy data.
- C. Implement Retrieval Augmented Generation (RAG) for in-context responses.
- D. Use pre-training and data augmentation on the company policy data.
Show the answer
Answer: C. RAG supplies policy content as retrieved context at inference time, which is the cheapest path to grounded answers and stays current as policies change. Retraining and pre-training are the most expensive options, and fine-tuning still requires a labeled dataset and a training job that RAG avoids.
Question 7. A company wants to learn about generative AI applications in an experimental environment. Which solution will meet this requirement MOST cost-effectively?
- A. Amazon Q Developer
- B. Amazon SageMaker JumpStart
- C. Amazon Bedrock PartyRock
- D. Amazon Q Business
Show the answer
Answer: C. PartyRock is a low-cost, hands-on playground built for learning and experimenting with generative AI without infrastructure setup, which fits an experimental environment most cost-effectively. Q Developer and Q Business are production assistants for coding and enterprise data. SageMaker JumpStart deploys models that incur compute charges, so PartyRock is the most cost-effective choice for learning.
Question 8. An ecommerce company wants to group customers based on their purchase history and preferences to personalize the user experience of the company's application. Which ML technique should the company use?
- A. Classification
- B. Clustering
- C. Regression
- D. Content generation
Show the answer
Answer: B. Clustering groups records by similarity without predefined labels, which fits segmenting customers by purchase history and preferences. Classification assigns records to known categories and regression predicts numeric values, neither of which is the goal here. Content generation creates new output such as text or images and does not group customers.
Question 9. A company wants to develop an educational game where users answer questions such as the following: "A jar contains six red, four green, and three yellow marbles. What is the probability of choosing a green marble from the jar?" Which solution meets these requirements with the LEAST operational overhead?
- A. Use supervised learning to create a regression model that will predict probability.
- B. Use reinforcement learning to train a model to return the probability.
- C. Use code that will calculate probability by using simple rules and computations.
- D. Use unsupervised learning to create a model that will estimate probability density.
Show the answer
Answer: C. The answer is an exact, deterministic calculation, so simple rule-based code is the lowest-overhead and most reliable solution. ML is not appropriate when a specific correct outcome is required rather than a prediction. Building supervised, reinforcement, or unsupervised models would add training data, tuning, and maintenance for a problem that basic arithmetic already solves.
Question 10. A financial company wants to flag all credit card activity as possibly fraudulent or non-fraudulent based on transaction data. Which type of ML model meets these requirements?
- A. Regression
- B. Diffusion
- C. Binary classification
- D. Multi-class classification
Show the answer
Answer: C. Sorting each transaction into one of two outcomes, fraudulent or non-fraudulent, is a binary classification problem. Regression predicts continuous numbers, and diffusion models generate images. Multi-class classification is for three or more categories, but this scenario has only two.
Unlock the full 30-question sampler
The PDF has all 30 questions plus an answer key, with the same written explanations the book uses, covering why the right answer is right.
What you get
The sampler is free. Here is exactly what lands in your inbox.
Print it or work it on screen, then score yourself against the answer key.
Nothing is written for marketing. Every question is lifted from the study guide itself.
After the PDF, a few short exam-technique emails over the next two weeks, including a 10% code for the full study guide. Unsubscribe any time.
The full AWS AI Practitioner study guide
The complete guide covers the AWS Certified AI Practitioner (AIF-C01) exam objective by objective in plain English, with the full practice question bank behind these samples.
From $24.99
Simply Certified is an independent publisher. AWS is a trademark of Amazon Web Services, Inc., which does not sponsor or endorse this product. This page and the sampler are independent study material and are not affiliated with, endorsed by, or sponsored by AWS.