AIF-C01 Question Bank · Latest 50 Questions Free Preview (Page 5/5)
Questions 41–50 of the latest 50 AIF-C01 questions (5 pages total), with answers and explanations, no login required. Practice all 435 questions online.
Question 41 · No. 426 · Single choice
A company runs an application on servers in an Amazon VPC. The company’s application uses Amazon Bedrock APIs for AI features. The company does not want API calls to travel across the public internet. Which solution will meet this requirement?
- A. Use AWS PrivateLink to establish a private connection between the Amazon VPC and Amazon Bedrock.
- B. Sign API requests by using an access key ID and a secret access key that is associated with an IAM principal.
- C. Move the application to an on-premises server. Make API calls to the public endpoint of Amazon Bedrock.
- D. Encrypt the data in transit from clients to the APIs that have been implemented in the Amazon VPC. Encrypt the data in transit for API calls from the Amazon VPC to Amazon Bedrock.
Answer: A
Explanation: AWS PrivateLink provides private connectivity between Amazon VPCs and AWS services like Amazon Bedrock, ensuring that API calls do not travel across the public internet. Option B (signing API requests) only authenticates the calls but does not prevent traffic from going over the public internet. Option C explicitly uses public endpoints, which would traverse the public internet. Option D (encryption in transit) secures the data but does not prevent the traffic from leaving the VPC environment over the public internet. Therefore, AWS PrivateLink is the correct solution to meet the requirement.
Question 42 · No. 427 · Single choice
A company is deploying a new AI application to generate content for internal users. Which strategy will make the application output more deterministic?
- A. Decreasing the temperature
- B. Increasing the learning rate
- C. Setting stop sequences
- D. Setting the token count
Answer: A
Explanation: Decreasing the temperature makes AI output more deterministic because temperature controls the randomness of token selection during generation. A lower temperature value (closer to 0) makes the model select the highest-probability tokens more consistently, resulting in more predictable and repeatable outputs. Option B (learning rate) is a training-time parameter that doesn't affect inference determinism. Option C (stop sequences) only controls when generation stops, not the randomness of the content. Option D (token count) only controls output length, not determinism.
Question 43 · No. 428 · Single choice
A company maintains a large product catalog with detailed descriptions. The company wants to build an AI assistant to answer customer questions about the products. However, the company’s labeled training data is limited. Which solution will meet these requirements with the LEAST implementation effort?
- A. Use a Retrieval Augmented Generation (RAG) architecture to query the product database at runtime. Provide relevant, unmodified product descriptions as context for the foundation model (FM).
- B. Fine-tune a foundation model (FM) on the restricted labeled data. Automatically refresh the model with novel product details weekly for accuracy.
- C. Deploy a foundation model (FM) for each product category. Implement a routing layer to direct customer queries to the appropriate specialized model.
- D. Create a new custom foundation model (FM) that is trained on the product database. Optimize the FM for minimal token usage during inference.
Answer: A
Explanation: Option A is the best choice because a Retrieval Augmented Generation (RAG) architecture requires the least implementation effort. It leverages the existing product catalog as a knowledge base and retrieves relevant information at query time without requiring any model training or fine-tuning. The other options (B, C, D) all involve significant implementation effort such as fine-tuning models, deploying multiple specialized models, or training a new custom foundation model from scratch. Given that the company has limited labeled training data, RAG is the most practical and efficient solution as it works with the existing product descriptions directly.
Question 44 · No. 429 · Single choice
A company uses foundation models (FMs) to automate daily tasks. An AI practitioner at the company is creating system instructions that include context that is relevant to the tasks. The AI practitioner wants to save and reuse the instructions in daily interactions with FMs in Amazon Bedrock. Which Amazon Bedrock solution will meet these requirements?
- A. Knowledge Bases
- B. Guardrails
- C. Playgrounds
- D. Prompt management
Answer: D
Explanation: Prompt management in Amazon Bedrock is specifically designed to help users create, save, version, and reuse prompts across different interactions with foundation models. It allows AI practitioners to store system instructions and context once and then reuse them in daily interactions, which directly matches the requirement described in the question. Knowledge Bases (A) are used for retrieving information from data sources to augment FM responses, not for managing system instructions. Guardrails (B) are safety controls for content filtering. Playgrounds (C) are testing environments for experimenting with models and prompts, but don't provide persistent storage and reuse capabilities for production use.
Question 45 · No. 430 · Single choice
What is an example of structured data?
- A. A file of text comments from an online forum
- B. A compilation of video files that contains news broadcasts
- C. A CSV file that consists of measurement data
- D. Transcribed conversations between call center agents and customers
Answer: C
Explanation: A CSV file containing measurement data is a classic example of structured data because CSV (Comma-Separated Values) files follow a defined format with rows and columns, making the data easily searchable, organized, and analyzable. Options A, B, and D represent unstructured data: text comments, video files, and transcribed conversations do not have a predefined data model or consistent format, which makes them harder to organize and analyze compared to structured data.
Question 46 · No. 431 · Single choice
A company has developed an AI assistant application for customers by using Amazon Bedrock Agents. The company notices that the agents require additional context to make informed decisions consistently. The agents are not accessing external services to complete customer travel booking requests. The company needs to ensure that the agents access additional context when necessary. Which solution will meet this requirement?
- A. Negative prompting
- B. Model Context Protocol (MCP)
- C. Zero-shot prompting
- D. Custom API to provide more context
Answer: B
Explanation: The question describes a scenario where Amazon Bedrock Agents are not accessing external services to complete customer travel booking requests and need additional context. Model Context Protocol (MCP) is specifically designed to enable AI agents to access external data sources and services in a standardized way. Amazon Bedrock Agents supports MCP to allow agents to retrieve additional context from external systems when needed to make informed decisions. Negative prompting and Zero-shot prompting are prompt engineering techniques that do not enable external service access. A Custom API would require additional development work and is not the standardized solution for this requirement.
Question 47 · No. 432 · Single choice
An AI Practitioner is using an LLM-as-a-judge in Amazon Bedrock to evaluate the quality of agent responses in a production environment. The AI practitioner wants to apply a built-in metric that assesses how thoroughly the agent responses address all parts of each prompt or question. Which metric will meet these requirements?
- A. Recall-Oriented Understudy for Gisting Evaluation (ROUGE)
- B. Completeness
- C. Following instructions
- D. Refusal
Answer: B
Explanation: The 'Completeness' metric in Amazon Bedrock's LLM-as-a-judge evaluation specifically assesses how thoroughly a model's response addresses all parts and aspects of the prompt or question. The question explicitly mentions 'how thoroughly the agent responses address all parts of each prompt,' which directly aligns with what the Completeness metric measures. ROUGE (A) is a general NLP metric for summarization quality based on n-gram overlap, not a built-in Bedrock metric for this purpose. 'Following instructions' (C) measures adherence to given instructions but not the comprehensiveness of addressing all prompt parts. 'Refusal' (D) evaluates whether the model appropriately declines certain requests, which is unrelated to thoroughness.
Question 48 · No. 433 · Single choice
A company is using supervised learning to train an AI model on a small labeled dataset that is specific to a target task. Which step of the foundation model (FM) lifecycle does this describe?
- A. Fine-tuning
- B. Data selection
- C. Pre-training
- D. Evaluation
Answer: A
Explanation: The question describes using supervised learning on a small, task-specific labeled dataset, which is the definition of fine-tuning in the foundation model lifecycle. Fine-tuning involves taking a pre-trained foundation model and further training it on a smaller dataset specific to the target task. Data selection (B) is about choosing data for pre-training, pre-training (C) involves large-scale training on massive datasets, and evaluation (D) is about assessing model performance, not training.
Question 49 · No. 434 · Single choice
A software development company wants to increase its speed to market delivery. The company needs a solution that helps developers generate real-time code and unit tests. Which AWS service meets these requirements?
- A. Amazon Personalize
- B. Amazon Q Developer
- C. Amazon Comprehend
- D. Amazon Kendra
Answer: B
Explanation: Amazon Q Developer is an AI-powered coding assistant that helps developers generate real-time code and unit tests, which directly addresses the company's need to increase speed to market delivery. Amazon Personalize is for recommendations, Amazon Comprehend is for NLP, and Amazon Kendra is for intelligent search.
Question 50 · No. 435 · Single choice
Which type of ML technique provides the MOST explainability?
- A. Linear regression
- B. Support vector machines
- C. Random cut forest (RCF)
- D. Neural network
Answer: A
Explanation: Linear regression is the most explainable ML technique because it uses a simple linear equation where each input feature has a coefficient that directly indicates its contribution to the output. This transparency allows users to easily understand how the model makes predictions. In contrast, support vector machines, random cut forest, and neural networks are more complex 'black box' models, making them harder to interpret.