30 Practice Questions with Answers and Explanations

Question 1

You are designing an Azure AI solution that must automatically scale during periods of unpredictable traffic.

Which hosting model is MOST appropriate?

A. Dedicated on-premises servers
B. Static virtual machines only
C. Manual scaling groups
D. Serverless compute

Answer

D. Serverless compute

Explanation

Serverless compute services such as Azure Functions automatically scale based on workload demand and reduce operational overhead.

Question 2

You need to generate semantic vector representations for documents used in a RAG solution.

Which service is MOST commonly used?

A. Azure Bastion
B. Azure OpenAI Service
C. Azure Firewall
D. Azure Route Server

Answer

B. Azure OpenAI Service

Explanation

Azure OpenAI Service provides embedding models that generate semantic vectors used in vector search and retrieval systems.

Question 3

You are building an AI application that retrieves enterprise documents using meaning rather than exact keyword matches.

Which capability should you implement?

A. Vector search
B. Blob replication
C. DNS forwarding
D. OCR masking

Answer

A. Vector search

Question 4

MULTIPLE ANSWER — Which components are commonly included in a hybrid search solution? (Choose THREE)

A. Vector retrieval
B. Keyword search
C. Semantic ranking
D. DHCP scopes
E. VPN tunneling

Answer

A. Vector retrieval
B. Keyword search
C. Semantic ranking

Question 5

You need to identify whether customer feedback expresses positive, neutral, or negative opinions.

Which capability should you use?

A. Object detection
B. OCR
C. Sentiment analysis
D. Image tagging

Answer

C. Sentiment analysis

Question 6

MATCHING — Match the capability to the correct scenario.

Options:

• Identify company names in contracts
• Create concise summaries of reports
• Extract text from scanned forms

Answer

Question 7

You need an AI agent to call APIs and execute workflows dynamically.

Which feature enables this capability?

A. Layout analysis
B. Semantic ranking
C. Function calling
D. Object segmentation

Answer

C. Function calling

Question 8

You are processing scanned PDFs that contain tables, headers, and forms.

Which capability is MOST important for preserving structure?

A. OCR only
B. Layout analysis
C. Translation
D. Speech synthesis

Answer

B. Layout analysis

Question 9

FILL IN THE BLANK

The process of splitting documents into smaller retrievable sections is called __________.

Answer

chunking

Question 10

You are implementing an AI search system that combines:

• Keyword matching
• Semantic similarity

What type of search is this?

A. Sequential search
B. OCR search
C. Relational search
D. Hybrid search

Answer

D. Hybrid search

Question 11

You need to extract structured invoice fields such as:

• Vendor name
• Total amount
• Invoice number

Which Azure service should you use?

A. Azure AI Document Intelligence
B. Azure Traffic Manager
C. Azure VPN Gateway
D. Azure DNS

Answer

A. Azure AI Document Intelligence

Question 12

You need to reduce hallucinations in a chatbot that answers questions from enterprise documentation.

Which approach should you implement?

A. Disable embeddings
B. Increase randomness
C. Grounding with RAG
D. Remove retrieval steps

Answer

C. Grounding with RAG

Question 13

HOTSPOT — Select the BEST capability for each requirement.

Options:

• Speech-to-text
• Object detection
• Key phrase extraction

Answer

Question 14

You are building a multimodal AI system.

Which inputs are examples of multimodal data? (Choose TWO)

A. Images
B. Audio
C. VLAN identifiers
D. Route tables

Answer

A. Images
B. Audio

Question 15

You need to monitor telemetry, logs, and performance metrics for Azure AI APIs.

Which Azure service should you use?

A. Azure Backup
B. Azure Monitor
C. Azure Front Door
D. Azure Bastion

Answer

B. Azure Monitor

Question 16

MULTIPLE ANSWER — Which techniques improve retrieval quality in RAG systems? (Choose THREE)

A. Semantic chunking
B. Metadata enrichment
C. Disabling semantic ranking
D. Removing indexes
E. Chunk overlap

Answer

A. Semantic chunking
B. Metadata enrichment
E. Chunk overlap

Question 17

You need to authenticate Azure resources securely without storing secrets in application code.

Which feature should you implement?

A. Shared admin passwords
B. Anonymous access
C. Managed identities
D. Public storage containers

Answer

C. Managed identities

Question 18

You are building a solution that analyzes handwritten forms.

Which capability is MOST important?

A. OCR with handwriting recognition
B. Image resizing
C. Object tracking
D. Speech synthesis

Answer

A. OCR with handwriting recognition

Question 19

DRAG AND DROP — Match the Azure service to its primary purpose.

Options:

• Workflow orchestration
• Image analysis
• Search and retrieval

Answer

Question 20

You need to create concise summaries of long legal contracts.

Which AI capability should you implement?

A. Translation
B. OCR
C. Text summarization
D. Face analysis

Answer

C. Text summarization

Question 21

You need to preserve headings, tables, and reading order for downstream AI reasoning.

Which output format is BEST suited?

A. Binary serialization
B. Markdown
C. JPEG
D. TIFF

Answer

B. Markdown

Question 22

You need an AI system to identify products, locations, and organizations in support tickets.

Which capability should you use?

A. OCR
B. Translation
C. Image segmentation
D. Named Entity Recognition

Answer

D. Named Entity Recognition

Question 23

MULTIPLE ANSWER — Which are benefits of vector embeddings? (Choose THREE)

A. Semantic similarity matching
B. Contextual retrieval
C. Natural language understanding
D. DNS optimization
E. Network segmentation

Answer

A. Semantic similarity matching
B. Contextual retrieval
C. Natural language understanding

Question 24

You need to orchestrate AI workflows between Azure services and third-party systems using a low-code approach.

Which service should you use?

A. Azure Load Balancer
B. Azure Logic Apps
C. Azure Firewall
D. Azure Site Recovery

Answer

B. Azure Logic Apps

Question 25

FILL IN THE BLANK

The process of supplying trusted external context to an LLM is called __________.

Answer

grounding

Question 26

You need an AI solution that retrieves only documents users are authorized to access.

Which approach is MOST appropriate?

A. Shared global administrator access
B. Anonymous indexes
C. Security trimming with RBAC
D. Public blob storage

Answer

C. Security trimming with RBAC

Question 27

You are building a computer vision solution that identifies multiple objects within warehouse images.

Which capability should you use?

A. OCR
B. Object detection
C. Translation
D. Sentiment analysis

Answer

B. Object detection

Question 28

You need to improve search precision by storing additional context such as:

• Department
• Region
• Document type

Which technique should you implement?

A. Metadata enrichment
B. Token suppression
C. DNS replication
D. Embedding deletion

Answer

A. Metadata enrichment

Question 29

You need to build a chatbot that retrieves current enterprise information before generating responses.

Which architecture is MOST appropriate?

A. Static FAQ architecture
B. RAG architecture
C. Traditional ETL architecture
D. Batch processing architecture

Answer

B. RAG architecture

Question 30

You are implementing enterprise AI governance and want to restrict access to sensitive retrieval indexes.

Which practice BEST improves security?

A. Disable authentication
B. Public anonymous access
C. RBAC enforcement
D. Shared credentials across teams

Answer

C. RBAC enforcement

Explanation

Role-based access control (RBAC) ensures that:

Enterprise governance policies are enforced

Users access only authorized resources

AI systems retrieve approved content

Go to the AI-103 Exam Prep Hub main page

30 Practice Questions with Answers and Explanations

Question 1

You are developing a chatbot that must answer questions using only approved internal company documents.

Which technique should you implement to reduce hallucinations?

A. Increasing temperature settings
B. Grounding with retrieval
C. Removing semantic ranking
D. Disabling vector search

Answer

B. Grounding with retrieval

Explanation

Grounding uses trusted enterprise data retrieved at runtime to provide accurate and context-aware responses.

Question 2

You need to analyze video footage to detect and classify objects such as forklifts and pallets.

Which capability should you use?

A. Named Entity Recognition
B. OCR
C. Object detection
D. Text summarization

Answer

C. Object detection

Question 3

A company wants to preserve document structure, headings, bullet lists, and tables for downstream AI reasoning.

Which output format is MOST appropriate?

A. TIFF
B. CSV
C. Binary encoding
D. Markdown

Answer

D. Markdown

Question 4

MULTIPLE ANSWER — Which are common stages in a RAG ingestion pipeline? (Choose THREE)

A. VPN configuration
B. Embedding generation
C. Vector indexing
D. Document chunking
E. DHCP reservation

Answer

B. Embedding generation
C. Vector indexing
D. Document chunking

Question 5

You need an AI system to identify customer emotions within support conversations.

Which capability should you implement?

A. Sentiment analysis
B. Image segmentation
C. OCR preprocessing
D. Face verification

Answer

A. Sentiment analysis

Question 6

MATCHING — Match the service to the correct functionality.

Options:

• Image analysis
• Semantic retrieval
• Generative AI and embeddings

Answer

Question 7

You are designing an AI solution that must authenticate securely between Azure services without storing credentials in code.

Which feature should you implement?

A. Shared administrator passwords
B. Public anonymous access
C. Managed identities
D. Embedded API keys in source control

Answer

C. Managed identities

Question 8

You need to retrieve semantically similar documents even when users do not use exact keywords.

Which search capability enables this?

A. DNS lookup
B. Vector search
C. Binary search
D. OCR indexing

Answer

B. Vector search

Question 9

FILL IN THE BLANK

The process of converting images of text into machine-readable text is called __________.

Answer

OCR

Question 10

You need an AI agent to dynamically execute workflows such as:

• Querying APIs
• Updating tickets
• Sending notifications

Which feature supports this requirement?

A. Function calling
B. Layout analysis
C. Object tracking
D. Translation

Answer

A. Function calling

Question 11

You are implementing a retrieval system that combines:

• Keyword search
• Vector similarity
• Semantic ranking

What type of search is this?

A. Lexical-only retrieval
B. Sequential search
C. Binary retrieval
D. Hybrid search

Answer

D. Hybrid search

Question 12

You need to extract:

• Vendor names
• Totals
• Invoice IDs

from scanned invoices.

Which Azure service is MOST appropriate?

A. Azure Firewall
B. Azure AI Document Intelligence
C. Azure DNS
D. Azure Virtual WAN

Answer

B. Azure AI Document Intelligence

Question 13

HOTSPOT — Select the BEST capability for each requirement.

Options:

• Speech-to-text
• Object detection
• Named Entity Recognition

Answer

Question 14

You need to monitor API latency, request volume, and failures in an Azure AI solution.

Which service should you use?

A. Azure Backup
B. Azure DNS
C. Azure Monitor
D. Azure Bastion

Answer

C. Azure Monitor

Question 15

MULTIPLE ANSWER — Which approaches commonly improve retrieval quality? (Choose THREE)

A. Semantic chunking
B. Metadata enrichment
C. Chunk overlap
D. Removing embeddings
E. Disabling ranking

Answer

A. Semantic chunking
B. Metadata enrichment
C. Chunk overlap

Question 16

You need to classify incoming support tickets into categories such as:

• Billing
• Technical issue
• Sales inquiry

Which capability should you use?

A. OCR
B. Text classification
C. Face recognition
D. Image tagging

Answer

B. Text classification

Question 17

You are building a multimodal AI pipeline.

Which data types are examples of multimodal input? (Choose TWO)

A. Images
B. DNS zones
C. Routing tables
D. Audio

Answer

A. Images
D. Audio

Question 18

You need to preserve reading order and table structure when extracting content from PDFs.

Which capability is MOST important?

A. Sentiment analysis
B. Layout analysis
C. Translation
D. Speech synthesis

Answer

B. Layout analysis

Question 19

DRAG AND DROP — Match the concept to its description.

Options:

• Providing trusted context to an LLM
• Splitting documents into smaller sections
• Vector representations of semantic meaning

Answer

Question 20

You need to summarize lengthy research reports automatically.

Which capability should you implement?

A. OCR masking
B. Image segmentation
C. Translation
D. Text summarization

Answer

D. Text summarization

Question 21

You are building a voice-enabled assistant that accepts spoken commands.

Which capability converts speech into text?

A. OCR
B. Speech-to-text
C. Image classification
D. Object segmentation

Answer

B. Speech-to-text

Question 22

FILL IN THE BLANK

A retrieval pipeline that uses external data to improve AI response accuracy is called __________-Augmented Generation.

Answer

Retrieval

Question 23

You need to improve search filtering by storing contextual information such as:

• Department
• Classification level
• Region

Which technique should you implement?

A. Token suppression
B. Metadata enrichment
C. Vector truncation
D. OCR masking

Answer

B. Metadata enrichment

Question 24

MULTIPLE ANSWER — Which are benefits of grounding AI responses? (Choose THREE)

A. Reduced hallucinations
B. Elimination of indexes
C. Better enterprise relevance
D. Improved factual accuracy
E. Removal of embeddings

Answer

A. Reduced hallucinations
C. Better enterprise relevance
D. Improved factual accuracy

Question 25

You are implementing an enterprise AI search solution that must enforce document-level security.

Which approach should you use?

A. Public anonymous indexes
B. Shared administrator accounts
C. Security trimming with RBAC
D. Disabled authentication

Answer

C. Security trimming with RBAC

Question 26

You need to orchestrate AI workflows between Azure services and external APIs using a low-code platform.

Which service should you use?

A. Azure Load Balancer
B. Azure Logic Apps
C. Azure Front Door
D. Azure Traffic Manager

Answer

B. Azure Logic Apps

Question 27

You are analyzing handwritten forms submitted by customers.

Which capability is MOST important?

A. Translation
B. Image compression
C. Speech synthesis
D. OCR with handwriting recognition

Answer

D. OCR with handwriting recognition

Question 28

You need to generate semantic vectors for similarity-based retrieval.

What are these vectors called?

A. Tokens
B. Classifiers
C. Entities
D. Embeddings

Answer

D. Embeddings

Question 29

You need to create an AI application that retrieves the latest enterprise content before generating responses.

Which architecture is MOST appropriate?

A. Batch ETL architecture
B. Static FAQ architecture
C. RAG architecture
D. Relational replication architecture

Answer

C. RAG architecture

Question 30

You are implementing enterprise AI governance and want to ensure users can only retrieve authorized documents.

Which practice BEST supports this requirement?

A. Shared credentials
B. Anonymous storage access
C. Public search indexes
D. Role-based access control (RBAC)

Answer

D. Role-based access control (RBAC)

Explanation

RBAC restricts access to authorized users and supports secure enterprise AI retrieval architectures.

Go to the AI-103 Exam Prep Hub main page

30 Practice Questions with Answers and Explanations

Question 1

You are deploying a generative AI application that must respond consistently with minimal randomness.

Which parameter should you LOWER?

A. Frequency penalty
B. Max tokens
C. Temperature
D. Top-p

Answer

C. Temperature

Explanation

Lower temperature values produce more deterministic and predictable outputs.

Question 2

You need to detect whether uploaded images contain inappropriate or unsafe content.

Which capability should you implement?

A. Content moderation
B. OCR
C. Named Entity Recognition
D. Translation

Answer

A. Content moderation

Question 3

You need to preserve semantic continuity between adjacent chunks in a retrieval pipeline.

Which technique should you use?

A. Metadata suppression
B. Chunk overlap
C. Token truncation
D. OCR masking

Answer

B. Chunk overlap

Question 4

MULTIPLE ANSWER — Which capabilities are commonly associated with Azure AI Search? (Choose THREE)

A. VPN tunneling
B. Semantic ranking
C. Hybrid search
D. DHCP management
E. Vector indexing

Answer

B. Semantic ranking
C. Hybrid search
E. Vector indexing

Question 5

You need to extract printed and handwritten text from scanned insurance forms.

Which service is MOST appropriate?

A. Azure AI Vision only
B. Azure AI Document Intelligence
C. Azure DNS
D. Azure Route Server

Answer

B. Azure AI Document Intelligence

Question 6

MATCHING — Match the concept to its description.

Options:

• Numeric semantic representations
• Providing trusted external context
• Searching by intent and meaning

Answer

Question 7

You need an AI system that can:

• Retrieve knowledge
• Use tools
• Execute actions
• Maintain conversational state

What type of architecture is this?

A. Static FAQ system
B. Batch ETL workflow
C. Relational reporting system
D. Agentic AI architecture

Answer

D. Agentic AI architecture

Question 8

You need to identify positive and negative sentiment in social media posts.

Which capability should you use?

A. OCR
B. Sentiment analysis
C. Image segmentation
D. Face detection

Answer

B. Sentiment analysis

Question 9

FILL IN THE BLANK

A vector representation of semantic meaning is called an __________.

Answer

embedding

Question 10

You need to identify products, organizations, and locations within customer emails.

Which capability should you implement?

A. Translation
B. Named Entity Recognition
C. OCR masking
D. Image tagging

Answer

B. Named Entity Recognition

Question 11

You need to securely authenticate between Azure resources without storing secrets.

Which feature should you use?

A. Managed identities
B. Shared passwords
C. Public access keys
D. Anonymous authentication

Answer

A. Managed identities

Question 12

You are implementing a chatbot that retrieves enterprise documents before generating responses.

Which architecture should you implement?

A. Static response architecture
B. Relational replication architecture
C. Retrieval-Augmented Generation (RAG)
D. Batch transformation architecture

Answer

C. Retrieval-Augmented Generation (RAG)

Question 13

HOTSPOT — Select the BEST capability for each requirement.

Options:

• Text summarization
• Speech-to-text
• Object detection

Answer

Question 14

You need to process:

• Audio
• Images
• Documents
• Video

What type of AI system is this?

A. Lexical AI system
B. Multimodal AI system
C. Relational AI system
D. Sequential AI system

Answer

B. Multimodal AI system

Question 15

MULTIPLE ANSWER — Which techniques commonly improve retrieval relevance? (Choose THREE)

A. Metadata enrichment
B. Disabling ranking
C. Semantic chunking
D. Removing embeddings
E. Hybrid retrieval

Answer

A. Metadata enrichment
C. Semantic chunking
E. Hybrid retrieval

Question 16

You need to preserve document reading order, headings, and tables during extraction.

Which capability is MOST important?

A. OCR only
B. Layout analysis
C. Sentiment analysis
D. Speech synthesis

Answer

B. Layout analysis

Question 17

You need to orchestrate workflows between Azure AI services and external APIs using a low-code solution.

Which Azure service should you use?

A. Azure Traffic Manager
B. Azure Firewall
C. Azure Logic Apps
D. Azure Bastion

Answer

C. Azure Logic Apps

Question 18

You are building a search solution that retrieves content using both keywords and semantic similarity.

What type of search is this?

A. Sequential search
B. OCR search
C. Hybrid search
D. Static indexing

Answer

C. Hybrid search

Question 19

DRAG AND DROP — Match the Azure service to its primary functionality.

Options:

• Search and vector retrieval
• Image analysis
• Generative AI models

Answer

Question 20

You need to monitor:

• API failures
• Latency
• Request throughput

Which Azure service should you use?

A. Azure Backup
B. Azure DNS
C. Azure Monitor
D. Azure Site Recovery

Answer

C. Azure Monitor

Question 21

You are building a solution that extracts invoice numbers and totals from PDFs.

Which Azure service should you use?

A. Azure AI Document Intelligence
B. Azure Front Door
C. Azure Virtual WAN
D. Azure Load Balancer

Answer

A. Azure AI Document Intelligence

Question 22

MULTIPLE ANSWER — Which are common benefits of grounding AI responses? (Choose THREE)

A. Reduced hallucinations
B. Improved factual accuracy
C. Better enterprise relevance
D. Elimination of chunking
E. Removal of embeddings

Answer

A. Reduced hallucinations
B. Improved factual accuracy
C. Better enterprise relevance

Question 23

You need an AI assistant to execute actions such as:

• Creating tickets
• Sending emails
• Calling APIs

Which feature enables this?

A. OCR preprocessing
B. Layout analysis
C. Image classification
D. Function calling

Answer

D. Function calling

Question 24

You need to automatically categorize support tickets into departments.

Which capability should you implement?

A. Text classification
B. Face recognition
C. Translation
D. Object tracking

Answer

A. Text classification

Question 25

FILL IN THE BLANK

The process of splitting large documents into smaller retrievable sections is called __________.

Answer

chunking

Question 26

You need to improve search filtering by storing attributes such as:

• Department
• Security level
• Region

Which technique should you implement?

A. OCR normalization
B. Token deletion
C. Metadata enrichment
D. Vector truncation

Answer

C. Metadata enrichment

Question 27

You need to build an AI application that retrieves semantically similar documents.

Which capability should you use?

A. DNS forwarding
B. Vector search
C. Blob replication
D. VPN routing

Answer

B. Vector search

Question 28

You are implementing an enterprise AI retrieval system that must enforce document-level permissions.

Which approach should you use?

A. Shared administrator accounts
B. Anonymous indexes
C. Public blob access
D. Security trimming with RBAC

Answer

D. Security trimming with RBAC

Question 29

You need to identify forklifts and pallets within warehouse images.

Which computer vision capability should you implement?

A. OCR
B. Translation
C. Object detection
D. Sentiment analysis

Answer

C. Object detection

Question 30

You need to generate semantic vectors used for retrieval pipelines.

Which Azure service is MOST commonly used?

A. Azure OpenAI Service
B. Azure DNS
C. Azure Firewall
D. Azure Backup

Answer

A. Azure OpenAI Service

Explanation

Azure OpenAI embedding models generate semantic vectors that support:

• Vector search
• Similarity matching
• Hybrid retrieval
• RAG pipelines

Go to the AI-103 Exam Prep Hub main page