AWS AI Practitioner AIF-C - Free Practice Quiz

Question 1 of 60

Which AWS service provides a fully managed environment for building generative AI applications using foundation models?

A AWS Glue B Amazon SageMaker Studio C Amazon Bedrock D Amazon Comprehend

1 / 60

Question 2 of 60

A company wants to add a conversational chatbot to its customer service website. Which AWS service is best suited for building this chatbot?

A Amazon Transcribe B Amazon Polly C Amazon Comprehend D Amazon Lex

2 / 60

Question 3 of 60

Which machine learning approach involves training a model using labeled data where the correct output is known?

A Reinforcement learning B Transfer learning C Supervised learning D Unsupervised learning

3 / 60

Question 4 of 60

A media company wants to automatically generate captions for its video content. Which AWS service should they use?

A Amazon Translate B Amazon Rekognition C Amazon Polly D Amazon Transcribe

4 / 60

Question 5 of 60

What is a foundation model?

A A large model pre-trained on broad data that can be adapted to many tasks B A small model trained on a single specific task C A model that only processes structured data D A rule-based AI system

5 / 60

Question 6 of 60

Which AWS custom silicon chip is designed specifically to accelerate machine learning training workloads?

A AWS Trainium B AWS Graviton C AWS Nitro D AWS Inferentia

6 / 60

Question 7 of 60

A retailer wants to provide personalized product recommendations to website visitors based on their browsing history. Which AWS service is most appropriate?

A Amazon Comprehend B Amazon Kendra C Amazon Rekognition D Amazon Personalize

7 / 60

Question 8 of 60

Which type of machine learning involves an agent learning to make decisions by receiving rewards or penalties?

A Unsupervised learning B Semi-supervised learning C Reinforcement learning D Supervised learning

8 / 60

Question 9 of 60

What is the term for when a large language model generates factually incorrect information that appears plausible?

A Underfitting B Hallucination C Bias D Overfitting

9 / 60

Question 10 of 60

Which AWS service provides pre-trained AI capabilities for detecting objects, faces, and text in images?

A Amazon Textract B Amazon Comprehend C Amazon Polly D Amazon Rekognition

10 / 60

Question 11 of 60

A company needs to translate customer support emails from Spanish to English automatically. Which AWS service should they use?

A Amazon Polly B Amazon Translate C Amazon Comprehend D Amazon Lex

11 / 60

Question 12 of 60

What does RAG stand for in the context of generative AI?

A Retrieval-Augmented Generation B Recursive Attention Grounding C Real-time AI Generation D Random Access Generation

12 / 60

Question 13 of 60

Which AWS service enables natural language search across enterprise data sources like SharePoint, S3, and databases?

A Amazon Lex B Amazon Textract C Amazon Comprehend D Amazon Kendra

13 / 60

Question 14 of 60

What is the primary purpose of Amazon Bedrock Guardrails?

A To accelerate model training B To implement safeguards that restrict harmful or undesirable model outputs C To improve model accuracy D To reduce inference costs

14 / 60

Question 15 of 60

Which prompt engineering technique provides the model with several examples of the desired input-output behavior before asking it to complete a new task?

A Zero-shot prompting B Chain-of-thought prompting C Few-shot prompting D System prompting

15 / 60

Question 16 of 60

A company wants to convert text responses from their AI chatbot into spoken audio for visually impaired users. Which AWS service should they use?

A Amazon Lex B Amazon Polly C Amazon Transcribe D Amazon Translate

16 / 60

Question 17 of 60

Which machine learning concept refers to a model performing well on training data but poorly on new, unseen data?

A Underfitting B Normalization C Regularization D Overfitting

17 / 60

Question 18 of 60

What type of AI model is specifically designed to generate new content such as text, images, or audio?

A Discriminative model B Generative model C Clustering model D Regression model

18 / 60

Question 19 of 60

Which AWS service provides a managed platform for building, training, and deploying machine learning models at scale?

A AWS Glue B Amazon EMR C Amazon SageMaker D Amazon Bedrock

19 / 60

Question 20 of 60

What is an embedding in the context of AI and machine learning?

A A numerical vector representation of data such as text or images B A type of neural network architecture C A method of compressing image files D A data encryption technique

20 / 60

Question 21 of 60

Which AWS chip is optimized for running inference workloads on trained machine learning models?

A AWS Graviton B AWS Inferentia C AWS Nitro D AWS Trainium

21 / 60

Question 22 of 60

A healthcare company wants to automatically extract patient names and dates from medical forms. What type of AI capability are they using?

A Image classification B Sentiment analysis C Named entity recognition D Speech synthesis

22 / 60

Question 23 of 60

Which of the following best describes unsupervised learning?

A Training a model with labeled input-output pairs B Training using a combination of labeled and unlabeled data C An agent learning through reward and punishment D Finding patterns in data without labeled examples

23 / 60

Question 24 of 60

What is the purpose of a vector database in a RAG architecture?

A To cache API responses for faster retrieval B To store raw training data for model fine-tuning C To store and search embeddings for semantic similarity retrieval D To manage model deployment configurations

24 / 60

Question 25 of 60

Which Amazon Bedrock feature allows you to ground model responses in your company's proprietary data sources?

A Bedrock Model Evaluation B Bedrock Knowledge Bases C Bedrock Fine-tuning D Bedrock Guardrails

25 / 60

Question 26 of 60

A company is concerned that its AI model for loan approvals might discriminate against certain demographic groups. This is an example of which AI risk?

A Model bias B Overfitting C Hallucination D Data leakage

26 / 60

Question 27 of 60

Which prompt engineering technique asks the model to reason through a problem step-by-step before giving a final answer?

A Chain-of-thought prompting B Few-shot prompting C Zero-shot prompting D System prompting

27 / 60

Question 28 of 60

What is the primary difference between fine-tuning a model and using RAG?

A Fine-tuning modifies model weights with new data, while RAG retrieves external data at inference time B RAG updates the model's weights, while fine-tuning uses external data C RAG requires labeled training data, while fine-tuning does not D Fine-tuning is cheaper than RAG

28 / 60

Question 29 of 60

Which AWS service provides NLP capabilities such as sentiment analysis, entity recognition, and topic modeling without requiring ML expertise?

A Amazon Translate B Amazon Comprehend C Amazon Kendra D Amazon Lex

29 / 60

Question 30 of 60

According to AWS's responsible AI principles, what is the concept of AI explainability?

A The ability to understand and explain how an AI model reaches its decisions B Ensuring AI systems are fully automated C Encrypting AI model outputs D Making AI models available to everyone

30 / 60

Question 31 of 60

A company wants to use Claude on Amazon Bedrock. Which company developed the Claude foundation models?

A Meta B Amazon C Anthropic D Google

31 / 60

Question 32 of 60

What is zero-shot prompting?

A Asking a model to perform a task without providing any examples B Prompting a model that has not been fine-tuned C Using the minimum number of tokens in a prompt D Prompting a model with no system instructions

32 / 60

Question 33 of 60

Which of the following is a key characteristic of large language models (LLMs)?

A They only work with a single language B They require labeled training data for every task C They can only process structured tabular data D They are trained on large amounts of text data and can generate coherent text

33 / 60

Question 34 of 60

A business analyst wants to understand how a sentiment analysis model classifies customer reviews. Which responsible AI concept is most relevant?

A Robustness B Fairness C Explainability D Privacy

34 / 60

Question 35 of 60

Which AWS service allows you to create, store, and manage custom terminologies for more accurate translations in specific domains?

A Amazon Translate B Amazon Lex C Amazon Kendra D Amazon Comprehend

35 / 60

Question 36 of 60

What does PII stand for in the context of AI data governance?

A Personally Identifiable Information B Protected Information Interface C Private Intelligence Input D Personal Inference Information

36 / 60

Question 37 of 60

A company wants to evaluate multiple foundation models on Amazon Bedrock to determine which best fits their use case. Which Bedrock feature should they use?

A Bedrock Knowledge Bases B Bedrock Fine-tuning C Bedrock Guardrails D Bedrock Model Evaluation

37 / 60

Question 38 of 60

Which of the following best describes the role of a model card?

A A hardware specification for AI accelerators B A type of prompt template C A document that provides transparency about a model's intended use, training data, and limitations D A pricing guide for foundation model API calls

38 / 60

Question 39 of 60

Which Amazon Titan model family is specifically designed for generating text embeddings?

A Amazon Titan Embeddings B Amazon Titan Text C Amazon Titan Multimodal D Amazon Titan Image Generator

39 / 60

Question 40 of 60

What is the primary risk of using a foundation model trained on internet data for a specialized enterprise use case without any customization?

A The model requires on-premises deployment B The model may lack domain-specific knowledge and produce inaccurate results C The model cannot be accessed via API D The model will be too slow for production use

40 / 60

Question 41 of 60

Which AWS service enables developers to build, deploy, and manage AI agents that can use tools and take multi-step actions?

A Amazon Lex B Amazon Bedrock Agents C Amazon Kendra D Amazon Comprehend

41 / 60

Question 42 of 60

A company processes millions of customer service calls. They want to detect when customers express frustration. Which AWS service provides this capability most directly?

A Amazon Transcribe B Amazon Polly C Amazon Comprehend D Amazon Translate

42 / 60

Question 43 of 60

What is the concept of 'grounding' in generative AI?

A Reducing the temperature parameter of a model B Connecting model outputs to verifiable, real-world information C Restricting the topics a model can discuss D Training a model on domain-specific data

43 / 60

Question 44 of 60

Which of the following is an example of a multimodal AI model?

A A model trained on multiple programming languages B A model deployed across multiple AWS regions C A model that processes both text and images as input D A model that only generates text

44 / 60

Question 45 of 60

According to AWS's responsible AI framework, which principle focuses on ensuring AI systems work as intended and can handle unexpected inputs?

A Privacy B Fairness C Explainability D Robustness

45 / 60

Question 46 of 60

A developer wants Llama models on Amazon Bedrock for a text generation application. Which company originally developed these models?

A Anthropic B Meta C Google D Amazon

46 / 60

Question 47 of 60

Which data practice is most important for preventing bias in AI models?

A Limiting training data to recent data only B Ensuring training data is representative, diverse, and properly labeled C Using only synthetic data for training D Using the largest possible dataset regardless of quality

47 / 60

Question 48 of 60

What is the function of Amazon Bedrock's content filters within Guardrails?

A To encrypt data sent to foundation models B To limit the number of API calls per minute C To block harmful, offensive, or inappropriate content in model inputs and outputs D To speed up model inference

48 / 60

Question 49 of 60

A company needs to comply with GDPR and wants to ensure its AI system does not store or expose user personal data. Which category of responsible AI does this primarily concern?

A Fairness B Explainability C Robustness D Privacy and data governance

49 / 60

Question 50 of 60

Which AWS service is used for converting documents with complex layouts, such as forms and tables, into machine-readable text?

A Amazon Comprehend B Amazon Transcribe C Amazon Rekognition D Amazon Textract

50 / 60

Question 51 of 60

In the context of Amazon Bedrock, what is 'model invocation'?

A Deploying a model to a production endpoint B The process of training a new foundation model C Calling a foundation model's API to generate a response D Fine-tuning a model with custom data

51 / 60

Question 52 of 60

Which of the following scenarios best illustrates responsible AI in action?

A A company that monitors its AI model for bias, tests for fairness, and documents model limitations B A company that deploys AI models without human oversight C An AI model that optimizes for profit above all other considerations D An AI system that maximizes engagement regardless of user wellbeing

52 / 60

Question 53 of 60

What is the primary advantage of using a managed AI service like Amazon Rekognition over building a custom computer vision model?

A Using a managed service eliminates the need for ML expertise and infrastructure management B Custom models cannot be deployed on AWS C Managed services are always cheaper than custom models D Managed services always achieve higher accuracy

53 / 60

Question 54 of 60

A startup wants to search for specific faces across a large video library for a media cataloging application. Which AWS service enables this?

A Amazon Comprehend B Amazon Kendra C Amazon Rekognition D Amazon Transcribe

54 / 60

Question 55 of 60

Which term describes the set of tokens a foundation model can process at one time, including both the input prompt and generated output?

A Attention head B Parameter count C Embedding dimension D Context window

55 / 60

Question 56 of 60

In machine learning, what is the purpose of a training dataset versus a validation dataset?

A Both datasets serve the same purpose B Training data is used to teach the model, validation data is used to tune hyperparameters and assess performance during training C Training data is public, validation data is private D Training data is used for testing, validation data is used for training

56 / 60

Question 57 of 60

Which AWS security service should be used to manage permissions for who can invoke foundation models on Amazon Bedrock?

A AWS Shield B Amazon Macie C Amazon GuardDuty D AWS IAM (Identity and Access Management)

57 / 60

Question 58 of 60

A company wants to ensure its generative AI application does not discuss competitor products. Which Amazon Bedrock feature can enforce this?

A Bedrock Model Evaluation B Bedrock Guardrails with denied topics C Bedrock Fine-tuning D Bedrock Knowledge Bases

58 / 60

Question 59 of 60

What is the key benefit of using AWS Trainium for machine learning workloads compared to general-purpose GPU instances?

A Trainium is designed for inference, not training B Trainium supports a wider range of ML frameworks C Trainium only supports Amazon's proprietary models D Trainium provides better price-performance for deep learning training workloads

59 / 60

Question 60 of 60

Which of the following best describes the concept of 'transfer learning'?

A Taking a model trained on one task and adapting it for a related task B Training a model entirely from scratch on new data C Moving a model from training to production D Transferring a model between AWS regions

60 / 60

AWS AI Practitioner AIF-C01 Practice Quiz