HOTSPOT
-
Select the correct prompt engineering technique from the following list for each description. Select each prompt engineering technique one time or not at all.
Explanation: Providing a small number of examples corresponds to few-shot prompting.
Requesting step-by-step reasoning corresponds to chain-of-thought prompting.
Performing a task with no examples corresponds to zero-shot prompting.
Question 61
A company is using a large language model (LLM) to create a generative AI assistant. The company must choose an AI technique to ensure that the AI assistant generates the most factually correct responses. The company selects the Retrieval Augmented Generation (RAG) technique.
Which limitation of LLMs is the company trying to reduce?
A. Hallucinations
B. Security
C. Nondeterminism
D. Interpretability
Show Answer
Correct Answer: A
Explanation: Retrieval Augmented Generation (RAG) reduces hallucinations by grounding the model’s responses in retrieved, authoritative documents, improving factual accuracy instead of relying solely on the model’s learned probabilities.
Question 62
An AI practitioner notices a large language model (LLM) is generating different responses for the same input across multiple invocations.
Which risk of AI does this describe?
A. Hallucinations
B. Nondeterminism
C. Accuracy
D. Multimodality
Show Answer
Correct Answer: B
Explanation: The scenario describes the same input producing different outputs across multiple runs, which is known as nondeterminism. Hallucinations involve false or fabricated content, accuracy concerns correctness, and multimodality refers to handling multiple input types, none of which explain variability across identical invocations.
Question 62
A company has trained a foundation model (FM) to perform a specific task. The company needs to fine-tune the FM to perform a different but related task.
Which fine-tuning method will meet this requirement?
A. Hyperparameter tuning
B. Pre-training
C. Transfer learning
D. Reinforcement learning
Show Answer
Correct Answer: C
Explanation: Transfer learning fine-tunes a pre-trained foundation model on a new but related task by reusing learned representations, which directly matches the requirement. Hyperparameter tuning adjusts training settings, pre-training trains from scratch, and reinforcement learning optimizes via rewards rather than supervised fine-tuning.
Question 63
A company wants to extract key insights from large policy documents to increase employee efficiency.
Which generative AI strategy meets this requirement?
A. Regression
B. Clustering
C. Summarization
D. Classification
Show Answer
Correct Answer: C
Explanation: The goal is to extract key insights from large policy documents so employees can understand them faster. Generative AI text summarization is specifically designed to condense long documents into concise summaries highlighting the most important points, unlike regression, clustering, or classification which serve different analytical purposes.
Question 63
What is the primary purpose of system prompts in generative AI applications?
A. To authenticate user credentials to access responses from AI
B. To generate random responses by AI
C. To define the role and behavioral boundaries of AI
D. To compress output data from AI
Show Answer
Correct Answer: C
Explanation: System prompts set the foundational instructions for a generative AI model, defining its role, persona, tone, goals, and behavioral constraints. They guide how the model should respond across all interactions, rather than handling authentication, randomness, or data compression.
Question 64
A company creates video content. The company wants to use generative AI to generate new creative content and to reduce video creation time.
Which solution will meet these requirements in the MOST operationally efficient way?
A. Use the Amazon Titan Image Generator model on Amazon Bedrock to generate intermediate images. Use video editing software to create videos.
B. Use the Amazon Nova Canvas model on Amazon Bedrock to generate intermediate images. Use video editing software to create videos.
C. Use the Amazon Nova Reel model on Amazon Bedrock to generate videos.
D. Use the Amazon Nova Pro model on Amazon Bedrock to generate videos.
Show Answer
Correct Answer: C
Explanation: The requirement is to generate new creative video content while minimizing operational effort and video creation time. Amazon Nova Reel on Amazon Bedrock is purpose-built for direct AI video generation, producing complete videos from prompts without needing intermediate image generation or external video editing workflows. This makes it more operationally efficient than image-only models (Titan Image Generator or Nova Canvas) and more suitable than Nova Pro, which is a general multimodal LLM rather than a video generation model.
Question 64
A company wants its AI models to be transparent and explainable.
Which combination of Amazon SageMaker AI features will meet these requirements? (Choose two.)
A. SageMaker Model Cards
B. SageMaker Pipelines
C. SageMaker Clarity
D. SageMaker Model Monitor
E. SageMaker Debugger
Show Answer
Correct Answer: A, C
Explanation: SageMaker Model Cards provide standardized documentation about a model’s intended use, training data, evaluation results, and limitations, which supports transparency. SageMaker Clarify provides bias detection and feature attribution (explainability) for models, helping explain predictions and assess fairness. Together, they address both transparency and explainability requirements.
Question 65
A company is introducing a new feature for its application. The feature will refine the style of output messages. The company will fine-tune a large language model (LLM) on Amazon Bedrock to implement the feature.
Which type of data does the company need to meet these requirements?
A. Samples of only input messages
B. Samples of only output messages
C. Samples of pairs of input and output messages
D. Separate samples of input and output messages
Show Answer
Correct Answer: C
Explanation: Fine-tuning an LLM to refine output style is a supervised learning task. The model must learn how given inputs should map to desired styled outputs, which requires paired examples of input messages and their corresponding output messages. Input-only, output-only, or unpaired datasets cannot teach this mapping.
Question 65
A company wants to customize Amazon Bedrock foundation models (FMs) to improve an application's performance. The company must prepare a training dataset for text-to-text model fine-tuning.
Which dataset format should the company use to train the models?
A. A JSON file with labeled data
B. A CSV file with unlabeled data
C. A CSV file with tabular data
D. A text file with unlabeled data
Show Answer
Correct Answer: A
Explanation: Amazon Bedrock text-to-text model fine-tuning requires a labeled training dataset where each record contains an input prompt and the expected output. This dataset is provided in JSON (specifically JSON Lines) format, making a JSON file with labeled data the correct choice.
$19
Get all 397 questions with detailed answers and explanations
Instant download HTML + PDF delivered the moment payment clears.
Secure Stripe checkout we never see or store your card details.
7-day refund if files are defective see our refund policy.
