AI-102 Practice Exam 62 — 20 Questions

Question 1

You are building a customer support chatbot using Azure OpenAI Service. The chatbot must only respond based on the company's product documentation and should not generate answers outside that scope. Which approach should you use?

Choose an answer to begin — your selection is scored in the full session.

Accepted Answer

Use Azure OpenAI On Your Data with a search index built from the documentation.. Azure OpenAI On Your Data allows you to ground the model's responses on a specific set of documents by connecting it to a search index (e.g., Azure Cognitive Search) built from the product documentation. This ensures the model retrieves relevant chunks from the index and generates answers solely based on that content, preventing out-of-scope responses without modifying the underlying model.

Answer

Implement content filters to block responses not found in the documentation.

Answer

Fine-tune a GPT-4 model on the product documentation.

Answer

Use prompt engineering with a system message instructing the model to only answer from the documentation.

Question 2

Your company uses Azure AI Language to process legal documents. You have built a custom NER model to extract parties, dates, and obligations. The model performs well on English documents but now you need to support French and German documents. You have no labeled data in those languages. You want to use the existing English model as a starting point. The solution must be cost-effective and avoid manual labeling as much as possible. You also need to ensure that the model can be retrained quickly when new document types are added. Which approach should you take?

Accepted Answer

Enable the multilingual option in the custom NER project and retrain the model using the existing English labeled data.. Enabling the multilingual option in custom NER allows training on the existing English labeled data, and the model can generalize to French and German without requiring labeled data in those languages. This is cost-effective and avoids manual labeling. Option A is wrong because using the prebuilt NER would not extract custom entities like parties, dates, and obligations in the specific format needed. Option C is wrong because translating documents introduces potential errors and latency, and may not capture legal nuances. Option D is wrong because building separate models requires translating the labeled data, which is additional effort and may lose context.

Answer

Use the prebuilt NER model for French and German to extract entities like dates and organizations, and then map them to your custom entity types.

Answer

Use the English custom NER model as-is and translate the French and German documents to English before processing.

Answer

Build separate custom NER models for French and German by translating the English labeled data using Azure AI Translator and training new models.

Question 3

You are reviewing the response from an Azure OpenAI Service chat completion API call. The finish_reason is 'stop'. What does this indicate?

Accepted Answer

The model completed the response naturally. The `finish_reason` field in the Azure OpenAI chat completion API response indicates why the model stopped generating tokens. A value of `'stop'` means the model encountered a natural stopping point, such as the end of a sentence or a logical conclusion, and completed the response without hitting any limits or filters.

Answer

The response was truncated because the token limit was reached

Answer

The response was blocked by the content filter

Answer

The model is still generating the response

Question 4

Refer to the exhibit. You are creating an Azure Cognitive Services account using an ARM template snippet. What type of account is being created?

Accepted Answer

Azure AI Services multi-service account. The ARM template snippet uses the 'CognitiveServices' resource type and sets 'kind' to 'CognitiveServices', which provisions a multi-service account that provides access to multiple Azure AI services (e.g., Language, Computer Vision, Translator) under a single endpoint and key. This is distinct from single-service accounts, which use specific 'kind' values like 'TextAnalytics' or 'ComputerVision'.

Answer

Azure AI Language

Answer

Azure AI Computer Vision

Answer

Azure OpenAI Service

Question 5

Your organization is migrating on-premises machine learning models to Azure. The models are used for real-time inference. You need to choose a service that provides managed endpoints with autoscaling and supports custom containers. Which service should you use?

Accepted Answer

Azure Machine Learning managed online endpoints. Azure Machine Learning managed online endpoints are the correct choice because they provide fully managed, autoscaling endpoints specifically designed for real-time inference. They support custom container images, allowing you to deploy any model packaged as a Docker container, and handle traffic splitting, health checks, and scaling automatically without managing underlying infrastructure.

Answer

Azure Functions

Answer

Azure Kubernetes Service (AKS) with manual scaling

Answer

Azure AI Services custom vision

Question 6

A company uses Azure Document Intelligence to extract data from invoices. They deploy the model to a container for on-premises processing. After deployment, they notice that the container consumes more memory than expected. What should they do to optimize memory usage?

Accepted Answer

Set the 'Memory' environment variable to a lower value in the container configuration. Azure Document Intelligence containers expose a 'Memory' environment variable that allows you to limit the container's memory allocation. By setting this variable to a lower value, you constrain the container's memory usage, which directly addresses the issue of higher-than-expected consumption. This is the recommended approach for optimizing memory in containerized deployments.

Answer

Use the 'Read' model instead of the 'Layout' model

Answer

Use the cloud API instead of the container

Answer

Reduce the batch size in the client application

Question 7

A financial services company uses Azure AI Document Intelligence to process loan applications. The solution extracts data from uploaded PDFs and stores it in Azure Cosmos DB. Recently, extraction errors increased due to poor-quality scans. The company needs to improve accuracy without manual intervention. The solution must also handle varying document layouts. You need to recommend a plan. What should you do?

Accepted Answer

Train a custom document extraction model using labeled samples of poor-quality scans and different layouts.. Training a custom document extraction model using labeled samples of poor-quality scans and varying layouts directly addresses the root cause of extraction errors. Azure AI Document Intelligence's custom model capability allows you to train on specific document types and quality issues, improving accuracy without manual intervention. This approach adapts to the company's need to handle diverse layouts and degraded image quality, which prebuilt models cannot reliably manage.

Answer

Increase the throughput (TPS) of the Document Intelligence resource.

Answer

Enable OCR enhancement in Document Intelligence.

Answer

Switch to Azure AI Language for entity extraction.

Question 8

You are deploying a Custom Vision model to an Azure Machine Learning managed endpoint using the above ARM template snippet. The deployment fails with a timeout error. Which parameter should you adjust?

Accepted Answer

Increase scoringTimeout to 'PT10M'. The deployment fails due to a timeout error, which indicates that the model inference is taking longer than the default scoring timeout (typically 5 seconds for Azure ML managed endpoints). Increasing the scoringTimeout parameter to 'PT10M' (ISO 8601 duration for 10 minutes) extends the allowed time for the endpoint to respond, resolving the timeout without altering compute or authentication settings.

Answer

Change model version to 2

Answer

Change compute to 'GPU'

Answer

Change authMode to 'AAD'

Question 9

Your company has a large collection of legal contracts in PDF format stored in Azure Blob Storage. You need to extract key clauses, parties, and effective dates using a custom model in Azure AI Document Intelligence. The model must be retrained monthly as new contract templates are added. What is the recommended approach to handle model versioning and retraining?

Accepted Answer

Train a new model version using the 'compose' operation or copy the existing model and retrain with new samples. Azure AI Document Intelligence supports model versioning through the 'compose' operation, which allows you to combine multiple trained models into a single composed model, or by copying an existing model and retraining it with new samples. This approach preserves the existing model's knowledge while incrementally updating it with new contract templates, avoiding the need to retrain from scratch each month.

Answer

Use a multi-model ensemble by training separate models per template

Answer

Retrain the model from scratch each month using all historical data

Answer

Use Azure Machine Learning pipelines to automate retraining and deploy a new endpoint

Question 10

Your company wants to build a custom Copilot for customer support using Microsoft Copilot Studio. The support team needs to query a backend CRM system securely using the Copilot. Which authentication method should you configure for the custom connector to the CRM?

Accepted Answer

Microsoft Entra ID OAuth 2.0. Microsoft Entra ID OAuth 2.0 is the correct authentication method because it provides secure, token-based delegated access to the backend CRM system without exposing long-lived credentials. Copilot Studio connectors that require user context and secure resource access should use OAuth 2.0 with Microsoft Entra ID, which supports modern authentication flows like authorization code grant and can enforce conditional access policies.

Answer

API key authentication

Answer

Basic authentication

Answer

Client certificate authentication

Question 11

You are designing a generative AI solution using Azure OpenAI Service. The solution must meet compliance requirements by preventing the model from generating sensitive personal data. Which TWO configurations should you implement? (Select TWO.)

Accepted Answer

Implement data loss prevention (DLP) policies using Microsoft Purview. Microsoft Purview DLP policies can scan and block sensitive data (e.g., PII, credit card numbers) in prompts and responses when integrated with Azure OpenAI Service, ensuring compliance by preventing data exfiltration. Azure AI Content Safety content filters allow you to configure severity thresholds to block harmful or sensitive content categories, directly preventing the model from generating personal data at the inference layer.

Answer

Add a system message instructing the model not to generate personal data

Answer

Deploy the model in a specific region

Answer

Fine-tune the model with a dataset that excludes personal data

Question 12

Which THREE actions should you take when designing a custom skill for an Azure AI Search enrichment pipeline? (Choose three.)

Accepted Answer

Define input and output parameters in the skillset definition. Custom skills in Azure AI Search require explicit input and output parameter definitions in the skillset JSON. These parameters map data from the enrichment pipeline into the skill and return results back, enabling the search service to pass context (e.g., document fields) and consume the skill's output for further indexing.

Answer

Ensure the skill executes within 2 minutes

Answer

Write the skill in a language supported by Azure AI Search

Question 13

A healthcare company is developing a solution to analyze patient feedback using Azure AI Language. The solution must extract key phrases, detect sentiment, and identify personally identifiable information (PII) such as patient names and medical record numbers from unstructured text. The company has strict compliance requirements: all text processing must occur within the United States region, and no data may leave the Azure geography. The development team has provisioned a Language resource in the East US region and has been testing the solution. During testing, the team notices that the PII detection feature is returning results, but the key phrase extraction and sentiment analysis are failing with a 403 error. The error message indicates that the resource is not allowed to access these features. The team has verified that the resource is in the S0 tier. What should the team do to resolve the issue?

Accepted Answer

Verify regional availability of the features and recreate the Language resource in a supported region if necessary.. The 403 error indicates an authorization or regional availability issue, not a tier or key problem. Key phrase extraction and sentiment analysis are available in all regions that support the Language service, but PII detection may have broader regional support. The team must verify that the East US region supports all three features; if not, they need to recreate the resource in a supported region like West US or South Central US to ensure compliance with the United States geography requirement.

Answer

Switch to the free tier to access all features.

Answer

Upgrade the Language resource to the S1 tier to enable all features.

Answer

Regenerate the API key and update the application code.

Question 14

You deploy the ARM template above to create an Azure AI Language resource. After deployment, you try to use the custom question answering feature but it is not available. What is the most likely reason?

Accepted Answer

The 'kind' property is set to 'TextAnalytics', which does not enable custom question answering.. The 'kind' property in the ARM template is set to 'TextAnalytics', which provisions a general-purpose Text Analytics resource. Custom question answering is a feature of the Azure AI Language service that requires the resource to be created with the 'kind' set to 'Language' or 'ConversationalLanguageUnderstanding' (depending on the API version). Because the resource is of the wrong kind, the custom question answering capability is not available, even though the deployment succeeds.

Answer

The apiVersion is too old to support custom question answering.

Answer

The SKU is set to S, but custom question answering requires a higher SKU.

Answer

The location does not support custom question answering.

Question 15

You are reviewing an index definition created with PowerShell. The index is used for a knowledge mining solution that extracts people and organizations from documents. Users report that when they type partial names in the search bar, the suggester does not return suggestions. What is the most likely reason?

Accepted Answer

The suggester sourceFields do not include people or organizations. A suggester in Azure Cognitive Search only returns suggestions for fields explicitly listed in its `sourceFields` property. If the `people` and `organizations` fields are not included in `sourceFields`, the suggester cannot match partial names typed in the search bar, even if those fields are indexed and searchable. The suggester relies on prefix matching against the specified source fields to generate suggestions.

Answer

The people and organizations fields should be Edm.String instead of Collection(Edm.String)

Answer

The id field is not defined as a key in the index

Answer

The suggester searchMode should be 'analyzingInfixMatching' which is incorrect

Question 16

A healthcare organization uses Azure Document Intelligence to process patient intake forms. They notice that the confidence scores for field extraction are low. What is the most likely cause?

Accepted Answer

The custom model was trained with only 10 labeled forms. Custom models in Azure Document Intelligence require a minimum of five labeled forms for training, but low confidence scores typically indicate insufficient training data. With only 10 labeled forms, the model lacks enough examples to generalize well across variations in handwriting, formatting, and field values, leading to poor extraction confidence.

Answer

The document resolution is too low

Answer

The document layout is not analyzed

Answer

The batch processing size is too large

Question 17

Refer to the exhibit. You run the Azure CLI command to list deleted cognitive services accounts. You need to recover the 'myTextAnalytics' account. Which command should you use?

Accepted Answer

az cognitiveservices account recover --name myTextAnalytics --resource-group <rg> --location eastus. The Azure CLI command `az cognitiveservices account recover` is specifically designed to restore a soft-deleted Cognitive Services account within the 90-day retention period. The command requires the account name, resource group, and location to identify the deleted resource for recovery.

Answer

az cognitiveservices account create --name myTextAnalytics --resource-group <rg> --kind TextAnalytics --location eastus

Answer

az cognitiveservices account restore --name myTextAnalytics --resource-group <rg> --location eastus

Answer

az cognitiveservices account undelete --name myTextAnalytics --resource-group <rg> --location eastus

Question 18

A company uses Azure Face API to detect faces in a crowd. They need to comply with GDPR and delete face data after 30 days. What should they implement?

Accepted Answer

Use the Face API Delete operation to remove stored face IDs.. The Azure Face API stores face data (e.g., face IDs) in a PersonGroup or FaceList, and GDPR compliance requires the ability to delete personal data on demand. The Face API provides a dedicated Delete operation (e.g., FaceList - Delete Face or PersonGroup Person - Delete) to remove stored face IDs and associated face data, enabling the 30-day deletion requirement without recreating the resource.

Answer

Enable data encryption at rest.

Answer

Set a retention policy on the Face API resource.

Answer

Recreate the Face API resource every 30 days.

Question 19

Your team uses Azure AI Language in a multi-region architecture. You need to ensure that the solution is resilient to regional outages. Which THREE actions should you take?

Accepted Answer

Use a global endpoint that routes to the nearest region. Azure AI Language global endpoints automatically route requests to the nearest available region, providing built-in failover without manual intervention. This ensures that if one region experiences an outage, traffic is redirected to the next closest healthy region, maintaining service continuity.

Answer

Disable regional endpoints

Answer

Use a single region with high availability enabled

Question 20

Your organization is deploying a generative AI chatbot using Azure OpenAI Service. The chatbot must answer questions based on internal documents stored in Azure Blob Storage. You need to implement a retrieval-augmented generation (RAG) solution. Which THREE components are required? (Select THREE.)

Accepted Answer

Azure AI Search index. Azure AI Search is the core indexing and retrieval engine in a RAG solution. It ingests documents from Azure Blob Storage, creates a searchable index, and enables vector or hybrid search to retrieve relevant chunks. The chatbot then uses these retrieved chunks as context for the Azure OpenAI model to generate grounded answers.

Answer

Azure Functions for preprocessing

Answer

Azure SQL Database for metadata