AI-102 Practice Test 20 — 15 Questions

Question 1

Refer to the exhibit. You are deploying a generative AI model as an online endpoint in Azure Machine Learning. You receive complaints that the endpoint returns 503 errors during peak hours. What is the most likely cause?

Accepted Answer

The manual scale setting with 2 instances may be insufficient for peak traffic.. Option A is correct because the scale type is Manual with 2 instances; during peak load, the number of instances may be insufficient, leading to 503 errors. Option B is wrong because requestTimeout of 30 seconds is reasonable. Option C is wrong because model version is specified. Option D is wrong because environment variables do not cause 503 errors directly.

Answer

The request timeout of 30 seconds is too short.

Answer

The environment variable MODEL_CACHE_SIZE is set too low.

Answer

The model version is not specified correctly.

Question 2

You are building a solution to generate product descriptions using Azure OpenAI Service. You need to ensure that the output adheres to a specific tone (professional, friendly) and length (50-100 words). Which parameter should you adjust?

Accepted Answer

Set the system message with instructions about tone and length.. Option B is correct because the system message defines the assistant's behavior and output characteristics. Option A is wrong because temperature affects creativity, not tone or length. Option C is wrong because top_p is for sampling diversity. Option D is wrong because max_tokens sets a hard limit but doesn't enforce a minimum or specific style.

Answer

Configure max_tokens to limit response length.

Answer

Modify the top_p parameter.

Answer

Adjust the temperature parameter.

Question 3

You are developing a generative AI application that must comply with responsible AI principles. Which Azure AI service should you use to detect and filter harmful content in both input prompts and output responses?

Accepted Answer

Azure AI Content Safety. Option D is correct because Azure AI Content Safety provides content filtering for harmful categories. Option A is wrong because Azure OpenAI Service provides the model, not content safety. Option B is wrong because Azure AI Language offers moderation APIs but Content Safety is dedicated. Option C is wrong because Microsoft Purview is for data governance, not content safety.

Answer

Microsoft Purview

Answer

Azure OpenAI Service

Answer

Azure AI Language

Question 4

You are developing a chat application that uses Azure OpenAI Service to answer customer queries. You need to ensure that the model does not generate responses containing internal company policies or confidential information. Which approach should you use?

Accepted Answer

Set the system message to instruct the model not to include confidential information.. Option B is correct because system messages provide instructions to the model on how to behave, including restricting certain topics. Option A is wrong because Azure AI Content Safety filters harmful content but does not prevent model from generating confidential info. Option C is wrong because fine-tuning may not guarantee avoidance of specific topics. Option D is wrong because embeddings are for retrieval, not for restricting model output.

Answer

Use Azure AI Search with index filtering to exclude documents with confidential info.

Answer

Fine-tune the model on a dataset that excludes confidential information.

Answer

Configure Azure AI Content Safety with custom categories for confidential terms.

Question 5

You need to generate realistic synthetic data using Azure OpenAI Service to train a machine learning model. The data must be diverse and cover edge cases. Which approach should you use?

Accepted Answer

Use prompt engineering with detailed instructions to generate varied examples.. Option A is correct because prompt engineering with detailed descriptions can guide the model to generate diverse data covering edge cases. Option B is wrong because fine-tuning on existing data may reduce diversity. Option C is wrong because temperature alone is not sufficient; prompt engineering is key. Option D is wrong because embeddings don't generate data.

Answer

Fine-tune the model on a small dataset of real examples.

Answer

Use Azure OpenAI embeddings to generate similar data points.

Answer

Set a high temperature parameter only.

Question 6

You are using Azure AI Document Intelligence to extract data from invoices. The invoices have varying layouts, and you need to build a custom model that can handle multiple document formats. Which type of model should you train?

Accepted Answer

Custom neural model. Option D is correct because a custom neural model is best for extracting data from documents with varied layouts. Option A is wrong because a prebuilt invoice model only works for fixed layouts. Option B is wrong because a custom template model requires consistent layouts. Option C is wrong because a custom composed model combines multiple custom models, but neural is better for varied layouts.

Answer

Custom template model

Answer

Prebuilt invoice model

Answer

Custom composed model

Question 7

Your company is using Azure OpenAI Service to generate marketing copy. The compliance team requires that all generated content be reviewed for sensitive data before delivery. You need to implement a solution that automatically scans the output for personally identifiable information (PII) and blocks it if detected. Which service should you integrate?

Accepted Answer

Azure AI Content Safety with a custom blocklist. Option A is correct because Azure AI Content Safety with a custom blocklist can flag and block PII in generated content. Option B is wrong because Microsoft Purview Information Protection is for data classification and labeling, not real-time scanning of generative output. Option C is wrong because Azure AI Language PII detection is for recognizing PII but not built for blocking in a pipeline. Option D is wrong because Microsoft Defender for Cloud Apps is for cloud app security, not content filtering.

Answer

Microsoft Purview Information Protection

Answer

Microsoft Defender for Cloud Apps

Answer

Azure AI Language PII detection

Question 8

You are deploying a conversational AI chatbot using Azure AI Language service. The chatbot must be able to switch between multiple intents in a single conversation without restarting the session. Which feature should you enable?

Accepted Answer

Active learning. Option B is correct because active learning allows the model to learn from user interactions and adapt to multi-intent scenarios. Option A is wrong because orchestration workflow is for routing to different skills, not for multi-intent within a single skill. Option C is wrong because prebuilt domain components are fixed and not adaptive. Option D is wrong because dynamic entity extraction is about entities, not intents.

Answer

Orchestration workflow

Answer

Dynamic entity extraction

Answer

Prebuilt domain components

Question 9

Your company is building a multi-tenant SaaS application using Azure AI Foundry. Each tenant (customer) should have isolated model deployments and data, but you want to share the base models across tenants to reduce costs. Which approach should you use?

Accepted Answer

Deploy one model per base model and use separate endpoints with routing logic per tenant.. Option D is correct because deploying a single model and using separate endpoints per tenant with dynamic routing enables isolation while sharing the base model. Option A is wrong because fine-tuned models per tenant increase costs. Option B is wrong because shared endpoints with tenant IDs in headers may cause data leakage. Option C is wrong because separate model deployments per tenant increases costs.

Answer

Use a single model deployment and a shared endpoint with tenant ID in the header.

Answer

Create separate fine-tuned models for each tenant.

Answer

Deploy separate base models for each tenant.

Question 10

You are building an agentic solution using Microsoft Semantic Kernel. The agent must autonomously decide when to call an external API to fetch real-time data. You want to minimize token usage and avoid unnecessary API calls. Which planner configuration should you use?

Accepted Answer

Use a SequentialPlanner with a stepwise strategy and explicit function parameter constraints. Option D is correct because a SequentialPlanner combined with a stepwise strategy ensures the agent only invokes the API when the context explicitly requires it, reducing token waste. Option A is incorrect because AutoInvoke with greedy planning may over-call APIs. Option B is incorrect because ManualInvoke requires user intervention, defeating autonomy. Option C is incorrect because ParallelPlanner may invoke multiple APIs concurrently, increasing token usage.

Answer

Use a ManualInvoke kernel with a sequential planner

Answer

Use a ParallelPlanner with a function-calling model

Answer

Use an AutoInvoke kernel with a greedy action planner

Question 11

You are implementing an agentic solution using Azure AI Agent Service. The agent needs to maintain conversation context across multiple turns. You configure the agent with a custom prompt that includes a 'system message' and 'few-shot examples'. However, after a few turns, the agent starts repeating the same responses. What is the most likely cause?

Accepted Answer

The 'max_context_length' is set too low, causing earlier turns to be truncated. Option B is correct because a low maximum context length truncates earlier conversation turns, causing the agent to lose context and repeat responses. Option A is incorrect because the token limit per request affects the length of individual responses, not repetition. Option C is incorrect because the Azure OpenAI API does not recycle context unless explicitly configured. Option D is incorrect because changing temperature affects randomness, not repetition due to context loss.

Answer

The 'max_tokens' parameter is set too low for the response

Answer

The agent is using a 'context recycling' feature that resets after each turn

Answer

The temperature is set too high, causing the model to become deterministic

Question 12

Your organization is deploying an agentic solution using Microsoft Copilot Studio. The agent must be able to escalate to a human agent when it cannot resolve a user's request. You need to ensure that the escalation includes the full conversation history. What should you configure?

Accepted Answer

Add a 'Transfer conversation' action and set it to include the full transcript. Option C is correct because 'Transfer conversation' automatically passes the full transcript to the human agent. Option A is incorrect because 'Create a ticket' does not transfer the conversation. Option B is incorrect because 'End conversation' terminates without escalation. Option D is incorrect because 'Start a new topic' does not include history.

Answer

Add an 'End conversation' action and configure a fallback

Answer

Add a 'Create a ticket' action in the topic

Answer

Add a 'Start a new topic' action with context variables

Question 13

You are designing an agentic solution using Azure AI Agent Service with a custom skill that calls an external REST API. The API has rate limits: 100 requests per minute per client. You need to ensure the agent respects this limit without degrading user experience. Which approach should you take?

Accepted Answer

Use a token bucket rate limiter with a shared counter stored in Azure Cache for Redis. Option D is correct because a token bucket with a shared counter across all agent instances ensures global rate limiting without dropping requests unexpectedly. Option A is incorrect because 'retry-after' headers require the agent to handle 429 responses, which can degrade UX. Option B is incorrect because exponential backoff can cause delays. Option C is incorrect because a fixed delay per request does not adapt to varying traffic.

Answer

Set a fixed delay of 600ms between each API call

Answer

Configure the skill to handle HTTP 429 responses with retry-after logic

Answer

Implement exponential backoff in the skill code

Question 14

You are building an agentic solution using Microsoft Semantic Kernel. The agent needs to orchestrate multiple plugins. One plugin returns a large dataset that exceeds the model's context window. What is the best way to handle this?

Accepted Answer

Use a summarization plugin to condense the data before passing it to the model. Option B is correct because summarizing the data reduces its size while preserving key information. Option A is incorrect because truncation may lose critical data. Option C is incorrect because splitting into multiple calls increases latency and tokens. Option D is incorrect because marking as sensitive does not solve the size issue.

Answer

Split the data into multiple smaller API calls and combine results

Answer

Truncate the data to fit the context window

Answer

Configure the plugin to return only a subset of the data and mark the rest as sensitive

Question 15

You are using Microsoft Copilot Studio to create an agent that helps users reset their passwords. The agent should first verify the user's identity using multi-factor authentication (MFA) before proceeding. Which feature should you configure?

Accepted Answer

Configure Authentication settings to require Microsoft Entra ID authentication with MFA policy. Option C is correct because 'Authentication' in Copilot Studio allows you to require Microsoft Entra ID sign-in with MFA. Option A is incorrect because 'Power Automate flow' can call MFA but is not the primary configuration. Option B is incorrect because 'Variable' does not enforce authentication. Option D is incorrect because 'Topic trigger' initiates the topic but does not enforce authentication.

Answer

Add a variable to store the user's identity status

Answer

Add a Power Automate flow that calls Microsoft Entra ID MFA

Answer

Use a 'Sign in' topic trigger from the customer channel

AI-102 Practice Test 20 — 15 Questions

Exhibit

AI-102 Practice Test 20 — 15 Questions

Exhibit