A company wants to build an ML model to detect abnormal patterns in sensor data. The company does not have labeled data for training.
Which ML method will meet these requirements?
A. Linear regression
B. Classification
C. Decision tree
D. Autoencoders
Show Answer
Correct Answer: D
Explanation: The task is anomaly detection on sensor data without labeled examples, which is an unsupervised learning problem. Autoencoders learn the normal data distribution by reconstructing inputs; abnormal patterns yield high reconstruction error, making them well-suited for unlabeled anomaly detection.
Question 21
Which option is a disadvantage of using generative AI models in production systems?
A. Possible high accuracy and reliability
B. Deterministic and consistent behavior
C. Negligible computational resource requirements
D. Hallucinations and inaccuracies
Show Answer
Correct Answer: D
Explanation: A key disadvantage of deploying generative AI in production is the risk of hallucinations and factual inaccuracies, where the model produces confident but incorrect outputs. The other options describe advantages or unrealistic properties (high accuracy, deterministic behavior, negligible compute), not disadvantages.
Question 22
A company is training ML models on datasets. The datasets contain some classes that have more examples than other classes. The company wants to measure how well the model balances detecting and labeling the classes.
Which metric should the company use?
A. Accuracy
B. Recall
C. Precision
D. F1 score
Show Answer
Correct Answer: D
Explanation: The dataset is imbalanced, so accuracy can be misleading. The company wants to evaluate both how well the model detects each class (recall) and how well it correctly labels predictions (precision). The F1 score combines precision and recall using their harmonic mean, making it well suited for assessing performance on imbalanced classification problems.
Question 22
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
Show Answer
Correct Answer: B
Explanation: The built-in Completeness metric in Amazon Bedrock evaluates whether an agent response fully addresses all parts of a given prompt or question. ROUGE measures textual overlap, Following instructions focuses on adherence to constraints, and Refusal checks whether the model appropriately declines to answer, so they do not meet the requirement.
Question 23
A financial company uses a generative AI model to assign credit limits to new customers. The company wants to make the decision-making process of the model more transparent to its customers.
Which solution meets these requirements?
A. Use a rule-based system instead of an ML model.
B. Apply explainable AI techniques to show customers which factors influenced the model’s decision.
C. Develop an interactive UI for customers and provide clear technical explanations about the system.
D. Increase the accuracy of the model to reduce the need for transparency.
Show Answer
Correct Answer: B
Explanation: The requirement is to make the model’s decision-making process transparent to customers. Explainable AI techniques (such as feature importance, local explanations, or reason codes) directly reveal which factors influenced a specific credit limit decision. The other options do not address transparency of an ML model’s decisions: replacing the model is unnecessary, a UI alone does not explain decisions, and higher accuracy does not improve transparency.
Question 23
A real estate company is developing an ML model to predict house prices by using sales and marketing data. The company wants to use feature engineering to build a model that makes accurate predictions.
Which approach will meet these requirements?
A. Understand patterns by providing data visualization.
B. Tune the model’s hyperparameters.
C. Create or select relevant features for model training.
D. Collect data from multiple sources.
Show Answer
Correct Answer: C
Explanation: Feature engineering focuses on creating, transforming, or selecting the most relevant features so the model can better learn the relationship between inputs and house prices. This directly improves predictive accuracy, whereas visualization, hyperparameter tuning, or data collection address different stages of the ML workflow.
Question 24
A company is using an Amazon Nova Canvas model to generate images. The model generates images successfully.
The company needs to prevent the model from including specific items in the generated images.
Which solution will meet this requirement?
A. Use a higher temperature value.
B. Use a more detailed prompt.
C. Use a negative prompt.
D. Use another foundation model (FM).
Show Answer
Correct Answer: C
Explanation: Image generation models such as Amazon Nova Canvas support negative prompts, which explicitly tell the model what elements to exclude from the generated image. Using a negative prompt is the correct way to prevent specific items from appearing, whereas temperature, prompt detail, or switching models do not directly enforce exclusions.
Question 24
A company stores its AI datasets in Amazon S3 buckets. The company wants to share the S3 buckets with its business partners. The company needs to avoid accidentally sharing sensitive data.
Which AWS service should the company use to discover sensitive data in the dataset?
A. Amazon Kendra
B. Amazon Macie
C. Amazon Textract
D. AWS Data Exchange
Show Answer
Correct Answer: B
Explanation: Amazon Macie uses machine learning to automatically discover, classify, and protect sensitive data (such as PII) stored in Amazon S3, helping prevent accidental data exposure when sharing buckets.
Question 25
HOTSPOT
-
Select the correct AWS service or tool from the following list for each use case. Select each AWS service or tool one time or not at all.
Show Answer
Correct Answer: Apply human feedback across the ML lifecycle:
Amazon SageMaker Ground Truth
Implement safeguards aligned with responsible AI policies:
Amazon Bedrock Guardrails
Detect potential bias during data preparation and model training:
Amazon SageMaker Clarify
Explanation: SageMaker Ground Truth enables human-in-the-loop labeling and feedback. Bedrock Guardrails provides safety and policy controls for responsible AI. SageMaker Clarify detects and explains bias in data and models.
Question 25
A company has deployed an AI application in production on AWS. The application’s responses have become less accurate over time.
The company needs a solution to send alerts when the application performance drifts.
Which AWS service or feature will meet this requirement?
A. Amazon Augmented AI (Amazon A2I)
B. Amazon SageMaker Model Monitor
C. Amazon Rekognition
D. AWS Trusted Advisor
Show Answer
Correct Answer: B
Explanation: The issue describes model performance degrading over time, which is typically caused by data or concept drift. Amazon SageMaker Model Monitor is specifically designed to continuously monitor models in production, detect data drift, concept drift, and prediction quality issues, and send alerts when thresholds are breached. The other options do not provide automated model performance drift detection.
$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.
