A developer trained a custom model on Amazon Bedrock using a training dataset that contains confidential data. The developer wants to ensure the custom model does not generate inference responses based on the confidential data. What should the developer do to prevent inference responses based on confidential data?

A. Delete the custom model. Remove the confidential data from the training dataset. Retrain the custom model.

B. Use dynamic data masking to mask confidential data in inference responses.

C. Use Amazon SageMaker to encrypt confidential data in inference responses.

D. Use AWS Key Management Service (AWS KMS) to encrypt confidential data within the custom model.

Which feature of Amazon OpenSearch Service enables companies to build vector database applications?

C. Scalable index management and approximate nearest neighbor search capabilities

A. Integration with Amazon S3 for object storage

B. Support for geospatial indexing and queries

C. Scalable index management and approximate nearest neighbor search capabilities

D. Ability to perform real-time analysis on streaming data

A company wants to display its best-selling products and their total sales across different retail locations over the past 12 months. Which AWS solution should the company use to automatically generate charts?

C. Amazon Q in Amazon QuickSight

A company developed an ML model for image classification. The company wants to deploy this model into production so a web application can use it. The company needs a solution to host the model and provide predictions without managing any underlying infrastructure. Which solution meets these requirements?

A. Deploy the model using Amazon SageMaker Serverless Inference.

B. Deploy the model using Amazon CloudFront.

C. Host the model and provide predictions using Amazon API Gateway.

D. Host the model and provide predictions using AWS Batch.

A company has petabytes of unlabeled customer data available for advertising campaigns. The company wants to segment customers into different tiers to target and promote its products. Which method should the company use to meet these requirements?

D. Reinforcement Learning from Human Feedback (RLHF)

A company performs quarterly forecasting to optimize inventory and meet anticipated demand. The company uses an ML model for forecasting. A developer is writing a report about the trained ML model to provide transparency and explainability to business stakeholders. To meet transparency and explainability requirements, which content should the developer include in the report?

B. Partial Dependence Plot (PDP)

C. Sample data used for training

Which option is a use case for generative AI models?

B. Creating realistic images from text descriptions for digital marketing

A. Using an intrusion detection system to improve network security

B. Creating realistic images from text descriptions for digital marketing

C. Improving database performance by optimizing indexes

D. Analyzing financial data to predict stock market trends

A company is designing a solution to generate protective eyewear images. Which solution meets these requirements?

A. Perform human-in-the-loop validation using Amazon SageMaker Ground Truth Plus

B. Use an Amazon Bedrock knowledge base for data augmentation

C. Use Amazon Rekognition for image recognition

D. Use Amazon QuickSight for data summarization

AWS AIF-C01 Chapter 1 Free Practice Test 2025 — AI Practitioner Exam Questions

AWS AIF-C01 Chapter 1 Practice Questions (100 Questions)

Practice questions for the AWS AIF-C01 (AI Practitioner) exam, Chapter 1.

Q1. A developer trained a custom model on Amazon Bedrock using a training dataset that contains confidential data. The developer wants to ensure the custom model does not generate inference responses based on the confidential data. What should the developer do to prevent inference responses based on confidential data?
- A. Delete the custom model. Remove the confidential data from the training dataset. Retrain the custom model.
- B. Use dynamic data masking to mask confidential data in inference responses.
- C. Use Amazon SageMaker to encrypt confidential data in inference responses.
- D. Use AWS Key Management Service (AWS KMS) to encrypt confidential data within the custom model.
View question →
Q2. Which feature of Amazon OpenSearch Service enables companies to build vector database applications?
- A. Integration with Amazon S3 for object storage
- B. Support for geospatial indexing and queries
- C. Scalable index management and approximate nearest neighbor search capabilities
- D. Ability to perform real-time analysis on streaming data
View question →
Q3. A company wants to display its best-selling products and their total sales across different retail locations over the past 12 months. Which AWS solution should the company use to automatically generate charts?
- A. Amazon Q on Amazon EC2
- B. Amazon Q Developer
- C. Amazon Q in Amazon QuickSight
- D. Amazon Q in AWS Chatbot
View question →
Q4. A company wants to create an interactive application for children that generates new stories based on classic tales. The company plans to use Amazon Bedrock and needs to ensure outputs and themes are appropriate for children. Which AWS service or feature satisfies these requirements?
- A. Amazon Rekognition
- B. Amazon Bedrock Playground
- C. Amazon Bedrock Guardrails
- D. Amazon Bedrock Agents
View question →
Q5. A company developed an ML model for image classification. The company wants to deploy this model into production so a web application can use it. The company needs a solution to host the model and provide predictions without managing any underlying infrastructure. Which solution meets these requirements?
- A. Deploy the model using Amazon SageMaker Serverless Inference.
- B. Deploy the model using Amazon CloudFront.
- C. Host the model and provide predictions using Amazon API Gateway.
- D. Host the model and provide predictions using AWS Batch.
View question →
Q6. A company has petabytes of unlabeled customer data available for advertising campaigns. The company wants to segment customers into different tiers to target and promote its products. Which method should the company use to meet these requirements?
- A. Supervised learning
- B. Unsupervised learning
- C. Reinforcement learning
- D. Reinforcement Learning from Human Feedback (RLHF)
View question →
Q7. A company performs quarterly forecasting to optimize inventory and meet anticipated demand. The company uses an ML model for forecasting. A developer is writing a report about the trained ML model to provide transparency and explainability to business stakeholders. To meet transparency and explainability requirements, which content should the developer include in the report?
- A. Model training code
- B. Partial Dependence Plot (PDP)
- C. Sample data used for training
- D. Model convergence table
View question →
Q8. Which option is a use case for generative AI models?
- A. Using an intrusion detection system to improve network security
- B. Creating realistic images from text descriptions for digital marketing
- C. Improving database performance by optimizing indexes
- D. Analyzing financial data to predict stock market trends
View question →
Q9. A developer is using a large language model (LLM) to generate content for marketing campaigns. The generated content appears coherent and factually plausible, yet it is incorrect. What issue has the LLM encountered?
- A. Data leakage
- B. Hallucination
- C. Overfitting
- D. Underfitting
View question →
Q10. A company is designing a solution to generate protective eyewear images. Which solution meets these requirements?
- A. Perform human-in-the-loop validation using Amazon SageMaker Ground Truth Plus
- B. Use an Amazon Bedrock knowledge base for data augmentation
- C. Use Amazon Rekognition for image recognition
- D. Use Amazon QuickSight for data summarization
View question →
Q11. A security-focused company is using Amazon Bedrock to deploy foundation models (FMs). The company wants to ensure only authorized users can access the models. It needs to identify unauthorized access attempts to configure appropriate AWS Identity and Access Management (IAM) policies and roles for FM access control. Which AWS service should the company use to identify access attempts to Amazon Bedrock?
- A. AWS Audit Manager
- B. AWS CloudTrail
- C. Amazon Fraud Detector
- D. AWS Trusted Advisor
View question →
Q12. A company manually reviews all resumes submitted in PDF format. As the company grows, the volume of resumes is expected to exceed its manual review capacity. The company needs an automated system to convert PDF resumes into plain text format for further processing. Which AWS service meets this requirement?
- A. Amazon Textract
- B. Amazon Personalize
- C. Amazon Lex
- D. Amazon Transcribe
View question →
Q13. A company wants to use a large language model (LLM) and Amazon Bedrock to build a chat interface for its product manuals, which are stored in PDF format. Which solution is the most cost-effective and efficient to meet these requirements?
- A. When a prompt is submitted to Amazon Bedrock, use prompt engineering to include a single PDF file as context in the user prompt.
- B. When a prompt is submitted to Amazon Bedrock, use prompt engineering to include all PDF files as context in the user prompt.
- C. Fine-tune the model using all PDF documents with Amazon Bedrock, then use the fine-tuned model to process user prompts.
- D. Upload the PDF documents to an Amazon Bedrock knowledge base. When users submit prompts to Amazon Bedrock, use the knowledge base to provide context.
View question →
Q14. Which term describes the numerical representation used by artificial intelligence and natural language processing (NLP) models to improve understanding of real-world objects and concepts in textual information?
- A. Embedding
- B. Token
- C. Model
- D. Binary file
View question →
Q15. A company is building an application that needs to generate synthetic data based on existing data. Which type of model can the company use to meet this requirement?
- A. Generative Adversarial Network (GAN)
- B. XGBoost
- C. Residual Neural Network
- D. WaveNet
View question →
Q16. A company wants to use generative AI to improve developer productivity and software development quality. It plans to adopt Amazon Q Developer. What can Amazon Q Developer do to help the company meet these goals?
- A. Generate code snippets, perform code reference tracing, and track open-source license compliance.
- B. Run applications without provisioning or managing servers.
- C. Enable voice commands for coding and provide natural language search.
- D. Use ML models to convert audio files into text documents.
View question →
Q17. A company wants to use Amazon Bedrock to build an application. The company has a limited budget and wants flexibility without long-term commitments. Which Amazon Bedrock pricing model meets these requirements?
- A. On-Demand
- B. Model Customization
- C. Provisioned Throughput
- D. Spot Instances
View question →
Q18. In generative AI models, what is a token?
- A. A token is the fundamental unit of input and output that generative AI models operate on, representing words, subwords, or other linguistic units.
- B. A pronoun is the mathematical representation of words or concepts used in generative AI models.
- C. A token is the pre-trained weight of a generative AI model fine-tuned for a specific task.
- D. A token is a specific prompt or instruction used by a generative AI model to generate output.
View question →
Q19. A financial institution wants to use Amazon Bedrock to develop a generative AI application. The application is hosted within a VPC. To meet regulatory compliance requirements, the VPC must not allow any internet traffic. Which AWS service or feature satisfies these requirements?
- A. AWS PrivateLink
- B. Amazon Macie
- C. Amazon CloudFront
- D. Internet Gateway
View question →
Q20. A company has built a deep learning model for object detection and deployed it into production. When the model analyzes images to identify objects, which generative AI process occurs?
- A. Training
- B. Inference
- C. Model Deployment
- D. Bias Correction
View question →
Q21. A company is using Amazon SageMaker Studio notebooks to build and train ML models. It stores data in an Amazon S3 bucket and needs to manage data flow from Amazon S3 into SageMaker Studio notebooks. Which solution satisfies this requirement?
- A. Use Amazon Inspector to monitor SageMaker Studio.
- B. Use Amazon Macie to monitor SageMaker Studio.
- C. Configure SageMaker to use an S3 VPC endpoint.
- D. Configure SageMaker to use S3 Glacier Deep Archive.
View question →
Q22. A company wants to use a domain-specific model. It wishes to avoid building a model from scratch. Instead, it wants to adapt a pre-trained model to create a model for its specific task. Which ML strategy satisfies these requirements?
- A. Increase the number of epochs.
- B. Use transfer learning.
- C. Decrease the number of epochs.
- D. Use unsupervised learning.
View question →
Q23. A company wants to use AI to protect its applications from threats. The AI solution must check whether IP addresses originate from suspicious sources. Which solution satisfies this requirement?
- A. Build a speech recognition system.
- B. Create a natural language processing (NLP) named entity recognition system.
- C. Develop an anomaly detection system.
- D. Create a fraud prediction system.
View question →
Q24. A company wants to use a large language model (LLM) to develop custom code. It needs to prevent the LLM from being manipulated by common prompt engineering techniques to perform harmful actions or expose sensitive information. Which action reduces these risks?
- A. Create a prompt template that teaches the LLM to detect attack patterns.
- B. Increase the temperature parameter in LLM invocation requests.
- C. Avoid using LLMs not listed in Amazon SageMaker.
- D. Reduce the number of input tokens when invoking the LLM.
View question →
Q25. A company is developing a model to predict the price of a specific product. The model performs well on the training dataset. However, when the company deploys the model into production, its performance significantly degrades. What should the company do to resolve this issue?
- A. Reduce the amount of data used during training.
- B. Add hyperparameters to the model.
- C. Increase the amount of data used during training.
- D. Increase the model training time.
View question →
Q26. A company wants to use a foundation model (FM) on Amazon Bedrock to build a chatbot. The FM needs to access encrypted data stored in an Amazon S3 bucket. The data is encrypted using Amazon S3 managed keys (SSE-S3). The FM fails when attempting to access the S3 bucket data. Which solution satisfies these requirements?
- A. Ensure the IAM role assumed by Amazon Bedrock has permissions to decrypt the data using the correct encryption key.
- B. Configure the S3 bucket’s access permissions to allow public access for internet-based access.
- C. Use prompt engineering techniques to instruct the model to locate information in Amazon S3.
- D. Ensure the S3 data does not contain sensitive information.
View question →
Q27. A company has a foundation model (FM) built using Amazon Bedrock to answer customer queries about its products. The company wants to validate the model’s responses to typical queries. It needs to upload a new dataset for Amazon Bedrock to use in validation. Which AWS service satisfies this requirement?
- A. Amazon S3
- B. Amazon Elastic Block Store (Amazon EBS)
- C. Amazon Elastic File System (Amazon EFS)
- D. AWS Snowcone
View question →
Q28. A company wants to evaluate the cost associated with inference generated by a large language model (LLM). The company plans to build a generative AI application using Amazon Bedrock. Which factor affects inference cost?
- A. Number of tokens consumed
- B. Temperature value
- C. Volume of data used to train the LLM
- D. Total training time
View question →
Q29. An AI company is working with an independent software vendor (ISV) to periodically assess its systems and processes. When the ISV’s compliance report becomes available, the company needs to receive an email notification. Which AWS service can satisfy this requirement?
- A. AWS Audit Manager
- B. AWS Artifact
- C. AWS Trusted Advisor
- D. AWS Data Exchange
View question →
Q30. A company wants to perform sentiment analysis using a large language model (LLM) on Amazon Bedrock. The company requires the LLM to produce consistent responses to identical input prompts. Which inference parameter adjustment satisfies this requirement?
- A. Decrease the temperature value
- B. Increase the temperature value
- C. Reduce the output token length
- D. Increase the maximum generation length
View question →
Q31. A company is implementing the Amazon Titan foundation model (FM) on Amazon Bedrock. It needs to augment the model with relevant data from its private data sources. Which solution satisfies this requirement?
- A. Use a different FM
- B. Select a lower temperature value
- C. Create an Amazon Bedrock Knowledge Base
- D. Enable model invocation logging
View question →
Q32. A company wants to develop a trivia game where users answer: 'A jar contains 6 red marbles, 4 green marbles, and 3 yellow marbles. What is the probability of selecting a green marble from the jar?' Which solution satisfies this requirement with minimal operational overhead?
- A. Use supervised learning to build a regression model that predicts probability.
- B. Use reinforcement learning to train a model that returns probability.
- C. Use code with simple rules and arithmetic to calculate the probability.
- D. Use unsupervised learning to build a model that estimates probability density.
View question →
Q33. Amazon SageMaker Clarify 提供曾些功能？
- A. 集成检索增强生成 (RAG) 工作流程
- B. 监控生产中 ML 模型的质量
- C. 记录有关 ML 模型的关键细节
- D. 识别数据准备过程中的潜在偏差
View question →
Q34. A company is building a machine learning (ML) model. The company collects data and analyzes it by creating correlation matrices, computing statistical summaries, and visualizing the data. Which stage of the ML pipeline is the company currently performing?
- A. Data preprocessing
- B. Feature engineering
- C. Exploratory data analysis
- D. Hyperparameter tuning
View question →
Q35. A company's documents are missing some words due to database errors. The company wants to build an ML model that can suggest plausible words to fill in the missing text. Which type of model satisfies this requirement?
- A. Topic modeling
- B. Clustering model
- C. Prescriptive ML model
- D. BERT-based model
View question →
Q36. A company is building a chatbot to improve user experience. It plans to use Amazon Bedrock's foundation model (FM) for intent detection. The company wants to improve intent detection accuracy using few-shot learning. To meet these requirements, what data does the company need?
- A. Paired chatbot responses and correct user intents
- B. Paired user messages and correct chatbot responses
- C. Paired user messages and correct user intents
- D. Paired user intents and correct chatbot responses
View question →
Q37. A company is building a large language model (LLM)-powered question-answering chatbot. The company aims to reduce the number of actions customer service agents must take when responding to customer inquiries. Which metric should the company use to evaluate the LLM chatbot's effectiveness?
- A. Website engagement rate
- B. Average call duration
- C. Corporate social responsibility
- D. Compliance
View question →
Q38. A company is using few-shot prompting on a hosted foundation model on Amazon Bedrock. The model currently uses 10 examples in its prompt. The model performs well with this configuration. The company wants to reduce monthly costs. Which solution meets these requirements?
- A. Fine-tune a custom model.
- B. Reduce the number of tokens in the prompt.
- C. Increase the number of prompts.
- D. Use 'provisioned throughput'.
View question →
Q39. A company has built an image classification model to predict plant diseases from photos of plant leaves. The company wants to evaluate how many images the model correctly classifies. Which evaluation metric should the company use to measure model performance?
- A. R-squared score
- B. Accuracy
- C. Root Mean Squared Error (RMSE)
- D. Learning rate
View question →
Q40. A large retail company receives thousands of product-related customer support queries daily. These queries require rapid handling and response. The company wants to implement Amazon Bedrock Agents. What key advantage does using Amazon Bedrock Agents provide for this retailer?
- A. Generate custom foundation models (FMs) to predict customer demand
- B. Automate repetitive tasks and orchestrate complex workflows
- C. Automatically invoke multiple foundation models (FMs) and aggregate results
- D. Select foundation models (FMs) based on predefined criteria and metrics
View question →
Q41. A company is training a foundation model (FM). The company wants to improve the model's accuracy to a specific target level. Which solution meets this requirement?
- A. Reduce batch size.
- B. Increase epochs.
- C. Decrease epochs.
- D. Increase temperature parameter.
View question →
Q42. A company has built a chatbot that answers natural language questions using images. The company wants to ensure the chatbot does not return inappropriate or unwanted images. Which solution meets these requirements?
- A. Implement a moderation API.
- B. Retrain the model using a general public dataset.
- C. Perform model validation.
- D. Automatically integrate user feedback.
View question →
Q43. A law firm wants to build an AI-powered application using a large language model (LLM) that reads legal documents and extracts key points. Which solution meets these requirements?
- A. Build an automated named entity recognition system.
- B. Create a recommendation engine.
- C. Develop a summarization chatbot.
- D. Develop a multilingual translation system.
View question →
Q44. A company wants to classify human genomic data into 20 categories based on genetic features. The company requires an ML algorithm whose internal mechanisms are interpretable to understand how they influence output results. Which ML algorithm meets these requirements?
- A. Decision tree
- B. Linear regression
- C. Logistic regression
- D. Neural network
View question →
Q45. A company wants to develop a large language model (LLM) application using Amazon Bedrock and customer data stored in Amazon S3. The company’s security policy mandates that each team can access only its own team’s customer data. Which solution meets these requirements?
- A. Create a custom Amazon Bedrock service role for each team, granting access only to that team’s customer data.
- B. Create a custom service role with Amazon S3 access permissions and require teams to specify the customer name in each Amazon Bedrock request.
- C. Re-encrypt personal data in Amazon S3 and update the S3 bucket policy to allow team access to customer data.
- D. Create an Amazon Bedrock role with full Amazon S3 access permissions and create an IAM role for each team that grants access only to that team’s customer folder.
View question →
Q46. A company wants to perform sentiment analysis using a large language model (LLM) on Amazon Bedrock. The company wants to know how much information a prompt can contain. Which factor should the company consider when making this decision?
- A. Temperature
- B. Context window
- C. Batch size
- D. Model size
View question →
Q47. An AI developer has built a deep learning model to classify material types in images. Now, the AI developer wants to measure the model’s performance. Which metric can help the AI developer evaluate the model’s performance?
- A. Confusion matrix
- B. Correlation matrix
- C. R² score
- D. Mean squared error (MSE)
View question →
Q48. An AI developer is building a model to generate diverse human-like images. The AI developer discovers bias in the input data, and this bias affects image generation and propagates into the model. Which technique can address this issue?
- A. Data augmentation for imbalanced classes
- B. Model monitoring for class distribution
- C. Retrieval-Augmented Generation (RAG)
- D. Image watermark detection
View question →
Q49. A company is building an ML model to analyze archival data. The company must run inference on large datasets—multiple gigabytes in size—and does not require immediate access to model prediction results. Which Amazon SageMaker inference option meets these requirements?
- A. Batch transform
- B. Real-time inference
- C. Serverless inference
- D. Asynchronous inference
View question →
Q50. A company needs to select a model from Amazon Bedrock for internal use. The company must ensure the model generates responses in a style preferred by its employees. Which approach should the company take to meet these requirements?
- A. Evaluate models using the built-in prompt dataset.
- B. Evaluate models using human reviewers and a custom prompt dataset.
- C. Identify models using public model leaderboards.
- D. Use the Model InvocationLatency runtime metric in Amazon CloudWatch when testing models.
View question →
Q51. A company is leveraging generative AI capabilities to assess its solution's security responsibilities across various implementation scopes. Based on these scopes, which solution scope assigns the greatest security responsibility to the company?
- A. Using a third-party enterprise application with embedded generative AI functionality.
- B. Building an application using an existing third-party generative AI foundation model (FM).
- C. Fine-tuning an existing third-party generative AI foundation model (FM) using business-specific data.
- D. Building and training a generative AI model from scratch using customer-owned, business-specific data.
View question →
Q52. A company uses Amazon SageMaker in its production environment for its ML pipeline. Its input data size reaches up to 1 GB, and inference time lasts up to 1 hour. The company requires near-real-time latency. Which SageMaker inference option satisfies these requirements?
- A. Real-time inference
- B. Serverless inference
- C. Asynchronous inference
- D. Batch transform
View question →
Q53. A company wants to deploy a language model to create an inference application on edge devices. The inference must achieve the lowest possible latency. Which solution satisfies these requirements?
- A. Deploying an optimized small language model (SLM) on the edge device.
- B. Deploying an optimized large language model (LLM) on the edge device.
- C. Integrating a centralized SLM API for asynchronous communication with the edge device.
- D. Integrating a centralized LLM API for asynchronous communication with the edge device.
View question →
Q54. A company wants to build an ML model using Amazon SageMaker. It needs to share and manage features used across multiple teams during model development. Which SageMaker feature satisfies these requirements?
- A. Amazon SageMaker Feature Store
- B. Amazon SageMaker Data Wrangler
- C. Amazon SageMaker Clarify
- D. Amazon SageMaker Model Cards
View question →
Q55. A company is building a chatbot for product recommendations using a pre-trained large language model (LLM). The company requires the LLM to output concise responses written in a specific language. Which solution ensures the LLM’s response quality meets the company’s expectations?
- A. Tuning the prompt.
- B. Selecting a different-sized LLM.
- C. Increasing temperature.
- D. Increasing top-k value.
View question →
Q56. A company uses Amazon Bedrock’s foundation models (FMs) for an AI-powered search tool. It aims to improve model accuracy by adapting the model using its own data. Which strategy successfully adapts the model?
- A. Providing labeled data with prompt and completion fields.
- B. Preparing the training dataset as a .txt file containing multiple lines in .csv format.
- C. Purchasing Amazon Bedrock’s Provisioned Throughput.
- D. Training the model on academic journals and textbooks.
View question →
Q57. An AI developer maintains an animal photo database. The developer wants to automatically identify and classify animals in photos under minimal human intervention. Which strategy satisfies these requirements?
- A. Object detection
- B. Anomaly detection
- C. Named entity recognition
- D. Inpainting
View question →
Q58. A research company has implemented a chatbot using foundation models (FMs) from Amazon Bedrock. The chatbot searches a database of academic research papers to answer user questions. After extensive prompt engineering attempts, the company realizes that the FM's performance is poor due to the highly complex scientific terminology in the research papers. What should the company do to improve the chatbot's performance?
- A. Use concise prompts to define how the FM should answer questions.
- B. Apply domain-adaptive fine-tuning to adapt the FM to complex scientific terminology.
- C. Modify the FM's inference parameters.
- D. Clean the research paper data by removing complex scientific terminology.
View question →
Q59. A healthcare company has deployed a disease detection model on Amazon Bedrock. To comply with privacy policies, the company wants to prevent the model from including patient personally identifiable information (PII) in its responses. Additionally, the company wants to receive notifications when policy violations occur. Which solution satisfies these requirements?
- A. Use Amazon Macie to scan model outputs for sensitive data and configure alerts for potential violations.
- B. Configure AWS CloudTrail to monitor model responses and create alerts for any detected PII.
- C. Use Amazon Bedrock’s Guardrails to filter content and configure Amazon CloudWatch alarms to notify on policy violations.
- D. Implement Amazon SageMaker Model Monitor to detect data drift and receive alerts when model quality degrades.
View question →
Q60. An educational institution is developing a question-and-answer application that uses generative AI foundation models to explain complex concepts. The institution wants the model’s response style to automatically adapt based on the asker’s profile (e.g., age group). Which solution meets these requirements with minimal implementation effort?
- A. Fine-tune the model using additional training data representing different age groups supported by the application.
- B. Include role descriptions in the prompt context to instruct the model to tailor responses for a specific age group.
- C. Use chain-of-thought reasoning to infer the appropriate response style and complexity for the user.
- D. Summarize response text based on user age to deliver shorter replies for younger users.
View question →
Q61. A social media company wants to use a large language model (LLM) for content moderation. The company wishes to evaluate LLM outputs to determine whether bias or latent discrimination against specific groups or individuals exists. Which data source should the company use to evaluate LLM outputs?
- A. User-generated content
- B. Moderation logs
- C. Content moderation guidelines
- D. Benchmark datasets
View question →
Q62. Which strategy evaluates the accuracy of a foundation model (FM) used for image classification tasks?
- A. Calculate the total resource cost incurred by the model.
- B. Measure model accuracy against a predefined benchmark dataset.
- C. Count the number of layers in the neural network.
- D. Assess color accuracy of images processed by the model.
View question →
Q63. A company maintains a multi-terabyte database for business analytics. It wants to build a generative AI application that generates SQL queries from natural-language input provided by employees who have minimal technical expertise. Which solution satisfies these requirements?
- A. Generative Pre-trained Transformer (GPT)
- B. Residual Neural Network
- C. Support Vector Machine
- D. WaveNet
View question →
Q64. Which metric measures the operational efficiency of a running generative AI model?
- A. Customer Satisfaction Score (CSAT)
- B. Training time per epoch
- C. Average response time
- D. Number of training instances
View question →
Q65. When fine-tuning a foundation model (FM), which option is a key benefit of continued pre-training?
- A. Helps reduce model complexity
- B. Improves model performance over time
- C. Reduces training time requirements
- D. Optimizes model inference time
View question →
Q66. A developer wants to use a foundation model (FM) to design a search application. Which type of foundation model should the developer use to support the search application?
- A. Multimodal embedding model
- B. Text embedding model
- C. Multimodal generative model
- D. Image generation model
View question →
Q67. A company is using Amazon Bedrock foundation models to summarize internal documents. The company has fine-tuned a custom model to improve summary quality. What action must the company take to use the custom model through Amazon Bedrock?
- A. Purchase Provisioned Throughput for the custom model.
- B. Deploy the custom model on an Amazon SageMaker endpoint for real-time inference.
- C. Register the model in the Amazon SageMaker Model Registry.
- D. Grant access to the custom model in Amazon Bedrock.
View question →
Q68. A company built a solution using generative AI. The solution uses a large language model (LLM) to translate training manuals from English into other languages. The company wants to assess the solution’s accuracy by reviewing the generated text. Which model evaluation strategy satisfies these requirements?
- A. Bilingual Evaluation Understudy (BLEU)
- B. Root Mean Square Error (RMSE)
- C. Recall-Oriented Understudy for Gisting Evaluation (ROUGE)
- D. F1 Score
View question →
Q69. How can enterprises securely use large language models (LLMs) on Amazon Bedrock?
- A. Design clear and specific prompts. Configure AWS Identity and Access Management (IAM) roles and policies.
- B. Enable AWS Audit Manager for automated model evaluation workflows.
- C. Enable Amazon Bedrock automated model evaluation jobs.
- D. Use Amazon CloudWatch Logs to make models interpretable and monitor bias.
View question →
Q70. A company is building a customer service chatbot. The company wants the chatbot to improve its responses by learning from past interactions and online resources. Which AI learning strategy provides this self-improvement capability?
- A. Supervised learning using a dataset of human-edited good and bad responses.
- B. Reinforcement learning that rewards positive customer feedback.
- C. Unsupervised learning to identify clusters of similar customer inquiries.
- D. Supervised learning using a frequently updated FAQ database.
View question →
Q71. A company wants to use a pre-trained generative AI model to generate marketing content. The company needs to ensure the generated content aligns with its brand voice and messaging requirements. Which solution satisfies these requirements?
- A. Optimize the model’s architecture and hyperparameters to improve overall performance.
- B. Increase the number of layers in the model architecture to enhance complexity.
- C. Create effective prompts that provide clear instructions and context to guide model generation.
- D. Select a large, diverse dataset to pre-train a new generative model.
View question →
Q72. A student is copying content generated by generative AI to write a paper. This situation represents which responsible AI challenge?
- A. Toxicity
- B. Hallucination
- C. Plagiarism
- D. Privacy
View question →
Q73. A company wants to build a chatbot to assist customers. The chatbot will help resolve technical issues and requires no human intervention. The company selected a foundation model (FM) for the chatbot. The chatbot must generate responses consistent with the company’s tone. Which solution satisfies these requirements?
- A. Set a low limit on the number of tokens generated by the FM.
- B. Use batch inference to process detailed responses.
- C. Iteratively test and refine prompts until the FM generates the desired responses.
- D. Define a high value for the temperature parameter.
View question →
Q74. A company has deployed surveillance cameras. The company uses a machine learning (ML) model to analyze images from the cameras to detect potential theft. The company discovers that the model disproportionately flags members of a specific racial group. What type of bias affects the model's output?
- A. Measurement bias
- B. Sampling bias
- C. Observer bias
- D. Confirmation bias
View question →
Q75. A company wants to use a foundation model (FM) on Amazon Bedrock for sentiment analysis. The company wishes to classify text passages as either positive or negative. Which prompt engineering strategy satisfies these requirements?
- A. Include example text passages with corresponding positive or negative labels in the prompt, then provide the new text passage to be classified.
- B. Provide a detailed explanation of sentiment analysis and how large language models (LLMs) work in the prompt.
- C. Provide only the new text passage to be classified, without any additional context or examples.
- D. Provide the new text passage along with examples of unrelated tasks, such as text summarization or question answering.
View question →
Q76. Which AWS service or feature enables development and research teams to rapidly deploy and use foundation models (FMs) within their own VPC?
- A. Amazon Personalize
- B. Amazon SageMaker JumpStart
- C. PartyRock in Amazon Bedrock Playground
- D. Amazon SageMaker endpoints
View question →
Q77. A company maintains a database containing petabytes of unstructured data from internal sources. The company wants to convert this data into structured formats to enable data scientists to perform machine learning (ML) tasks. Which service satisfies these requirements?
- A. Amazon Lex
- B. Amazon Rekognition
- C. Amazon Kinesis Data Streams
- D. AWS Glue
View question →
Q78. A company receives thousands of customer support interactions daily and wants to analyze them to identify common issues and derive insights. Which AWS service can meet this requirement?
- A. Amazon Lex
- B. Amazon Comprehend
- C. Amazon Transcribe
- D. Amazon Translate
View question →
Q79. A company built a generative AI solution using large language models (LLMs) to translate training manuals from English into other languages. The company wants to assess the solution’s accuracy by reviewing the generated translated text. Which model evaluation strategy satisfies this requirement?
- A. Bilingual Evaluation Understudy (BLEU)
- B. Root Mean Square Error (RMSE)
- C. Recall-Oriented Understudy for Gisting Evaluation (ROUGE)
- D. F1 score
View question →
Q80. A company manually reviews all submitted PDF-format resumes. As the company grows, the volume of resumes is expected to exceed its review capacity. Therefore, the company needs an automated system to convert PDF resumes into plain text format for further processing. Which AWS service satisfies this requirement?
- A. Amazon Textract
- B. Amazon Personalize
- C. Amazon Lex
- D. Amazon Transcribe
View question →
Q81. A company built a generative AI solution using large language models (LLMs) to translate training manuals from English into other languages. The company wants to assess the solution’s accuracy by reviewing the generated translated text. Which model evaluation strategy satisfies this requirement?
- A. Bilingual Evaluation Understudy (BLEU)
- B. Root Mean Square Error (RMSE)
- C. Recall-Oriented Understudy for Gisting Evaluation (ROUGE)
- D. F1 score
View question →
Q82. A company wants to use a language model to create an application for inference on edge devices. The inference must have the lowest possible latency. Which solution meets these requirements?
- A. Deploy an optimized small language model (SLM) on the edge device.
- B. Deploy an optimized large language model (LLM) on the edge device.
- C. Integrate a centralized small language model (SLM) API for asynchronous communication with the edge device.
- D. Integrate a centralized large language model (LLM) API for asynchronous communication with the edge device.
View question →
Q83. A company has built an image classification model to predict plant diseases from photos of plant leaves. The company wants to evaluate how many images the model correctly classified. Which evaluation metric should the company use to measure model performance?
- A. R-squared score
- B. Accuracy
- C. Root Mean Square Error (RMSE)
- D. Learning rate
View question →
Q84. A company is using Amazon Bedrock foundation models for internal document summarization use cases. The company has trained a custom model to improve summarization quality. What action must the company take to use the custom model via Amazon Bedrock?
- A. Purchase provisioned throughput for the custom model.
- B. Deploy the custom model in an Amazon SageMaker endpoint to enable real-time inference.
- C. Register the model in the Amazon SageMaker Model Registry.
- D. Grant permissions to access the custom model in Amazon Bedrock.
View question →
Q85. A company handles thousands of customer support interactions and wants to analyze them to identify common questions and derive insights. Which AWS service can the company use to meet this requirement?
- A. Amazon Lex
- B. Amazon Comprehend
- C. Amazon Transcribe
- D. Amazon Translate
View question →
Q86. A company is developing an LLM-based question-and-answer chatbot to reduce the number of actions call center agents must take when answering customer questions. Which business metric should the company use to evaluate the LLM chatbot’s effectiveness?
- A. Website engagement rate
- B. Average call duration
- C. Corporate social responsibility
- D. Regulatory compliance
View question →
Q87. A company is using Amazon Bedrock to implement the Amazon Titan foundation model (FM). The company needs to augment the model with relevant data from its private data sources. Which solution meets this requirement?
- A. Use a different foundation model (FM).
- B. Select a lower temperature value.
- C. Create an Amazon Bedrock knowledge base.
- D. Enable model invocation logging.
View question →
Q88. A digital device company wants to forecast customer demand for memory hardware. The company lacks coding experience or machine learning expertise and needs to develop a data-driven forecasting model. It must analyze both internal and external data. Which solution meets these requirements?
- A. Store data in Amazon S3. Use Amazon SageMaker built-in algorithms with data from Amazon S3 to build a machine learning model and perform demand forecasting.
- B. Import data into Amazon SageMaker Data Wrangler. Use SageMaker built-in algorithms to build a machine learning model and perform demand forecasting.
- C. Import data into Amazon SageMaker Data Wrangler. Use the Amazon Personalize Trending-Now template to build a machine learning model and perform demand forecasting.
- D. Import data into Amazon SageMaker Canvas. Build a machine learning model and perform demand forecasting by selecting values from the data within SageMaker Canvas.
View question →
Q89. Which metric can measure the efficiency of an AI model?
- A. Customer Satisfaction Score (CSAT)
- B. Training time per iteration cycle
- C. Average response time
- D. Number of training instances
View question →
Q90. A company wants to use a foundation model (FM) on Amazon Bedrock to develop a chatbot. The foundation model needs to access encrypted data stored in an Amazon S3 bucket. The data is encrypted using Amazon S3 managed keys (SSE-S3). When the FM attempts to access the S3 bucket data, it encounters a failure. Which solution meets these requirements?
- A. Ensure the IAM role assumed by Amazon Bedrock has permissions to use the correct encryption key to decrypt the data.
- B. Configure the S3 bucket’s access permissions to allow public access for internet-based access.
- C. Use prompt engineering techniques to instruct the model to locate information in Amazon S3.
- D. Ensure the S3 data does not contain sensitive information.
View question →
Q91. A company needs to select a model from Amazon Bedrock for internal use. The company must ensure that the model generates responses whose style aligns with employee preferences. What should the company do to meet these requirements?
- A. Evaluate the models using built-in prompt datasets.
- B. Evaluate the models using human reviewers and custom prompt datasets.
- C. Identify models using public model leaderboards.
- D. Use the ModelInvocationLatency runtime metric in Amazon CloudWatch when testing models.
View question →
Q92. A company is training a foundation model (FM). The company wants to improve the model’s accuracy to a specified acceptable level. Which solution meets these requirements?
- A. Decrease the batch size.
- B. Increase the number of iterations.
- C. Decrease the number of iterations.
- D. Increase the temperature parameter.
View question →
Q93. An AI practitioner wants to use an Amazon Bedrock foundation model to summarize customer service team chat transcripts. The practitioner wants to store invocation logs to monitor the model’s input and output data. Which strategy should the AI practitioner adopt?
- A. Configure AWS CloudTrail as the logging destination for the model.
- B. Enable invocation logging in Amazon Bedrock.
- C. Configure AWS Audit Manager as the logging destination for the model.
- D. Configure model invocation logging in Amazon EventBridge.
View question →
Q94. A company wants to use generative AI to improve developer productivity and software development efficiency. The company plans to use Amazon Q Developer. What can Amazon Q Developer do to help the company meet these requirements?
- A. Generate software code snippets, perform reference tracing, and track open-source license compliance.
- B. Run applications without configuring or managing servers.
- C. Enable voice commands for coding and provide natural language search.
- D. Use machine learning models to convert audio files into text documents.
View question →
Q95. An AI practitioner wants to use a foundation model (FM) to design a search application. The search application must support queries containing both text and images. Which type of feature matching (FM) should the AI practitioner use to support this search application?
- A. Multimodal embedding model
- B. Text embedding model
- C. Multimodal generative model
- D. Image generation model
View question →
Q96. A student at a university is copying content generated by generative AI to write a research paper. Which challenge related to responsible generative AI does this scenario represent?
- A. Toxicity
- B. Hallucination
- C. Plagiarism
- D. Privacy
View question →
Q97. A company wants to evaluate the inference cost associated with using a large language model (LLM); the company plans to use Amazon Bedrock. Which factor drives inference cost upward?
- A. Number of tokens consumed
- B. Temperature value
- C. Volume of data used to train the LLM
- D. Total training time
View question →
Q98. A company has developed a machine learning model for image classification. The company wants to deploy this model into a production environment so that a web application can use it for predictions. The company needs to implement a solution to host the model and provide prediction services without managing any underlying infrastructure. Which solution meets these requirements?
- A. Use Amazon SageMaker Serverless Inference to deploy the model.
- B. Use Amazon CloudFront to deploy the model.
- C. Use Amazon API Gateway to host the model and provide prediction services.
- D. Use AWS Batch to host the model and provide prediction services.
View question →
Q99. A research company has developed a chatbot using a foundation model (FM) from Amazon Bedrock. This chatbot can search for answers to questions from a large database of research papers. After extensive prompt engineering experiments, the company finds that the FM performs poorly because research papers contain complex scientific terminology. What should the company do to improve the chatbot’s performance?
- A. Use few-shot prompting to define how the FM should answer questions.
- B. Use domain adaptation fine-tuning to adapt the FM to complex scientific terminology.
- C. Change the FM inference parameters.
- D. Clean the research paper data to remove complex scientific terminology.
View question →
Q100. A company has installed surveillance cameras. It uses a machine learning model to analyze surveillance video footage to detect potential theft behavior. However, the company discovers that the model disproportionately misclassifies members of a particular racial group. Which type of bias is affecting the model’s output?
- A. Measurement bias
- B. Sampling bias
- C. Observer bias
- D. Confirmation bias
View question →