AWS AI Practitioner - Practice Test 1

Question 1

What is the fundamental difference between Generative AI and traditional (discriminative) Machine Learning?

Accepted Answer

Generative AI creates new content (text, images) similar to training data, while traditional ML predicts labels or values.

Answer

There is no difference, they are interchangeable terms.

Answer

Traditional ML requires GPUs, while Generative AI only runs on CPUs.

Answer

Generative AI classifies existing data, while traditional ML creates new data.

Question 2

A company wants to use foundation models (FMs) from leading providers like Anthropic and Cohere without managing the underlying infrastructure. Which AWS service is most suitable?

Accepted Answer

Amazon Bedrock

Answer

Amazon SageMaker JumpStart

Answer

AWS Lambda

Answer

Amazon Q

Question 3

What are TWO key pillars or principles of Responsible AI? (Select TWO)

Accepted Answer

Answer

Token maximization

Answer

Minimum cost

Answer

Inference speed

Answer

Fairness

Answer

Privacy and Security

Question 4

Which technique allows an LLM to answer questions using private and up-to-date corporate information, reducing hallucinations?

Accepted Answer

Retrieval-Augmented Generation (RAG)

Answer

Fine-tuning

Answer

Reinforcement Learning

Answer

Continued Pre-training

Question 5

Which Amazon SageMaker tool helps detect bias in data before training and in models after training?

Accepted Answer

SageMaker Edge Manager

Answer

SageMaker Pipelines

Answer

SageMaker Ground Truth

Answer

SageMaker Clarify

Question 6

What are TWO key capabilities of 'Agents for Amazon Bedrock'? (Select TWO)

Accepted Answer

Answer

Train foundation models from scratch

Answer

Break down complex tasks into logical steps (orchestration)

Answer

Create images from text

Answer

Execute actions by calling external APIs

Answer

Manage EC2 clusters for training

Question 7

If you want an AI model to generate more creative and varied responses, which inference parameter should you increase?

Accepted Answer

Temperature

Answer

Max Tokens

Answer

Stop Sequences

Answer

Top-K (set to 0)

Question 8

Which Amazon Bedrock feature is used to block harmful content and filter Personally Identifiable Information (PII) in model interactions?

Accepted Answer

Guardrails for Amazon Bedrock

Answer

Provisioned Throughput

Answer

Model Evaluation

Answer

Knowledge Bases

Question 9

Which AWS chip is specifically optimized to provide the lowest cost per inference for deep learning models?

Accepted Answer

AWS Nitro

Answer

AWS Trainium

Answer

AWS Graviton

Answer

AWS Inferentia

Question 10

What are TWO data sources natively supported by 'Knowledge Bases for Amazon Bedrock' for document ingestion? (Select TWO)

Accepted Answer

Answer

Amazon Glacier Deep Archive (directly)

Answer

Confluence (via connector)

Answer

Any local MySQL database

Answer

Physical paper files

Answer

Amazon S3