AI-102 Practice Exam 52 — 20 Questions

Question 1

You are a solution architect at a news agency. The agency publishes thousands of articles daily. You need to build a knowledge mining solution that enables journalists to search for articles by topic, sentiment, key people, and locations mentioned. The articles are stored as HTML files in Azure Blob Storage. The solution must also provide a summary for each article. You plan to use Azure AI Search with cognitive skills and Azure OpenAI. Which combination of skills and features should you include to meet all requirements with the best performance and accuracy?

Accepted Answer

Skillset with Entity Recognition skill, Sentiment skill, and Key Phrase Extraction skill. Use Azure OpenAI service to generate summaries via a custom skill that calls the GPT model. Enable semantic search.. Option C is correct because it uses Entity Recognition for people/locations, Sentiment for sentiment, Key Phrase Extraction for topics, and Azure OpenAI for summarization (via a custom skill). Option A uses Translator unnecessarily. Option B uses Text Analytics for health (not appropriate) and lacks summarization. Option D uses Document Intelligence which is for documents, not HTML.

Answer

Use Azure AI Document Intelligence to extract content from HTML, then use Azure AI Language to extract entities and sentiment. Index in Azure AI Search with semantic search.

Answer

Skillset with Entity Recognition skill, Sentiment skill, Key Phrase Extraction skill, and Text Translation skill. Enable semantic search.

Answer

Skillset with Entity Recognition skill, Sentiment skill, and Text Analytics for Health skill to extract medical terms. Use Azure OpenAI for summarization as a custom skill.

Question 2

A company is building a knowledge mining solution using Azure AI Search. They need to extract entities from a large set of PDF documents stored in Azure Blob Storage. The solution must use a built-in AI skill to identify people, organizations, and locations. Which TWO actions should be taken? (Choose two.)

Accepted Answer

Enable OCR (Optical Character Recognition) in the indexer configuration.. Option A is correct because the Entity Recognition skill identifies people, organizations, and locations. Option D is correct because the built-in OCR skill is needed to extract text from PDFs before entity recognition can occur. Options B, C, and E are either unrelated or unnecessary.

Answer

Configure the index to use a custom analyzer.

Answer

Add the Key Phrase Extraction skill to the skillset.

Answer

Deploy a custom skill using Azure Functions to extract entities.

Question 3

A retail company uses Azure Computer Vision to analyze customer traffic in stores. They process images from security cameras using the OCR API to detect product labels. Recently, the OCR accuracy has decreased for images with poor lighting. Which pre-processing step should the company implement to improve OCR accuracy?

Accepted Answer

Adjust brightness and contrast of images using image processing.. Option C is correct because poor lighting directly reduces the contrast between text and background, which is critical for OCR accuracy. Adjusting brightness and contrast improves the signal-to-noise ratio of the text regions, making character edges more distinct for the Azure Computer Vision OCR engine. This pre-processing step compensates for the lighting deficiency without altering the fundamental image content that the API relies on.

Answer

Convert images to grayscale before sending to OCR API.

Answer

Increase the image resolution before calling OCR API.

Answer

Reduce image size to decrease noise.

Question 4

Refer to the exhibit. An Azure Cognitive Services Computer Vision API call for image captioning is returning only one caption. The developer wants to get three possible captions ranked by confidence. Which parameter should be modified in the request?

Accepted Answer

Set the maxCandidates value to 3.. The `maxCandidates` parameter in the Computer Vision Image Analysis API controls the maximum number of captions returned in the response. By default, this value is 1, so only the top-ranked caption is returned. Setting `maxCandidates=3` instructs the API to return up to three captions, each with its own confidence score, ranked from highest to lowest confidence.

Answer

Use a different API version, such as 2023-04-01.

Answer

Modify the URL to point to a different image.

Answer

Change the language parameter to 'multi'.

Question 5

A company uses Azure Face API to verify employee identities for building access. They need to ensure that only live faces are used, not photos or videos. Which feature should they enable?

Accepted Answer

Enable liveness detection using session-based verification.. Option C is correct because Azure Face API's liveness detection with session-based verification is specifically designed to prevent spoofing attacks using photos, videos, or masks. It analyzes subtle cues such as micro-movements, texture, and depth to confirm the presence of a live person, ensuring that only live faces are accepted for identity verification.

Answer

Set a high confidence threshold for face matching.

Answer

Face identification with a large person group.

Answer

Face detection with attributes such as age and emotion.

Question 6

A retail company uses Azure Computer Vision to analyze customer traffic in stores. They deploy a custom object detection model to count customers and detect occupancy. After deployment, the model consistently underestimates the number of customers during peak hours. The company has retrained the model with more data but the issue persists. What is the most likely cause?

Accepted Answer

The training data does not adequately represent peak-hour scenarios.. The model consistently underestimates customer counts during peak hours, which indicates a distribution shift between the training data and the inference environment. Even after retraining with more data, the issue persists because the additional data likely still lacks sufficient representation of peak-hour scenarios (e.g., high density, occlusion, rapid movement). In Azure Custom Vision, object detection models learn from labeled examples; if the training set does not include diverse peak-hour images with varied lighting, crowd densities, and angles, the model will fail to generalize to those conditions.

Answer

The model is not being batch-processed for inference.

Answer

The model is overfitting to the training data.

Answer

The Computer Vision API version is outdated.

Question 7

A company uses Azure Custom Vision to build a classifier for defect detection on a manufacturing line. They have labeled images of products with and without defects. Which TWO actions should they take to improve model performance?

Accepted Answer

Use images with balanced numbers of defect and non-defect samples.. Option B is correct because balanced datasets prevent the model from becoming biased toward the majority class (e.g., non-defect images), which is critical for defect detection where defects are rare. Azure Custom Vision uses a weighted loss function during training, and class imbalance can cause the model to predict the majority class for most inputs, reducing recall for defects. Balanced samples ensure the model learns discriminative features for both classes equally.

Answer

Train for more iterations without validation.

Answer

Set the learning rate manually using the Custom Vision API.

Answer

Reduce the number of images per tag to avoid overfitting.

Question 8

A developer is building an application to extract text from scanned invoices using Azure Computer Vision's Read API. The invoices contain a mix of printed and handwritten text. The developer needs to ensure the highest accuracy for both types. Which parameter should they set in the API call?

Accepted Answer

No special parameter; the Read API automatically handles both.. The Read API in Azure Computer Vision is designed to extract text from images and documents, and it automatically handles both printed and handwritten text without requiring any special parameter. Setting the 'language' parameter to 'en' is optional and only improves accuracy for language-specific text, but it does not enable or disable handwriting recognition. Therefore, no additional parameter is needed to achieve the highest accuracy for both types.

Answer

Set the 'language' parameter to 'en' for English handwriting.

Answer

Specify the 'model-version' as '2022-04-30'

Answer

Use the 'mode' parameter set to 'Handwriting'

Question 9

A hospital uses Azure Custom Vision to classify X-ray images as normal or abnormal. The model achieves 98% accuracy on the test set. However, during deployment, the model misclassifies many abnormal cases as normal, causing missed diagnoses. The hospital has a class imbalance where abnormal cases are only 5% of the data. What should the data scientist do first to address this?

Accepted Answer

Use oversampling or class-weight techniques to balance the training.. Option D is correct because the primary issue is class imbalance, where abnormal cases constitute only 5% of the data. Oversampling (e.g., SMOTE) or class-weight techniques adjust the training process to give more importance to the minority class, directly addressing the model's bias toward the majority class and reducing false negatives. This is a standard preprocessing step in Custom Vision and other ML frameworks before tuning hyperparameters or changing algorithms.

Answer

Increase the number of training epochs.

Answer

Add more normal X-ray images to the dataset.

Answer

Switch to a different object detection algorithm.

Question 10

Which THREE actions can be performed using the Azure Custom Vision service?

Accepted Answer

Export a trained model to ONNX format for offline inference.. Option B is correct because Azure Custom Vision allows you to export trained models to ONNX format for offline inference. This enables running the model on edge devices or in environments without continuous internet connectivity, leveraging the ONNX runtime for efficient deployment.

Answer

Extract text from scanned receipts.

Answer

Identify specific individuals in a crowd using facial recognition.

Question 11

Drag and drop the steps to configure a multi-region disaster recovery for Azure Cognitive Services into the correct order.

Question 12

Which THREE factors should you consider when selecting a model for a generative AI solution on Azure?

Accepted Answer

Cost per token and deployment options.. Option A is correct because cost per token directly impacts operational expenses in pay-per-use models like GPT-4 on Azure OpenAI, and deployment options (e.g., global vs. regional, provisioned throughput) affect availability and scalability. These are fundamental considerations when operationalizing a generative AI solution.

Answer

Number of transformer layers in the model.

Answer

Training data source and licensing.

Question 13

A developer wants to use Azure OpenAI to generate text from a prompt. Which parameter controls the diversity of the generated output?

Accepted Answer

temperature. Temperature is the parameter that directly controls the randomness or diversity of the generated output by scaling the logits before applying the softmax function. A higher temperature (e.g., 1.0) increases the probability of less likely tokens, producing more creative and varied responses, while a lower temperature (e.g., 0.1) makes the output more deterministic and focused.

Answer

presence_penalty

Answer

frequency_penalty

Answer

max_tokens

Question 14

A company is using Azure OpenAI to generate customer support responses. They want to ensure the model does not use any personally identifiable information (PII) in its outputs. What should they implement?

Accepted Answer

Use Azure AI Content Safety to filter PII from the output.. Azure AI Content Safety provides built-in PII detection and redaction capabilities that can automatically scan and filter sensitive information from model outputs. This is the most reliable approach because it operates as a post-processing filter, catching PII that the model might generate despite instructions. Fine-tuning, prompt engineering, and system messages are all fallible because they rely on the model's compliance rather than enforced filtering.

Answer

Fine-tune the model on anonymized data.

Answer

Use prompt engineering to instruct the model to redact PII.

Answer

Use a system message instructing the model to avoid PII.

Question 15

A research lab wants to use Azure OpenAI to generate synthetic data for training a model. They need to generate a large volume of data quickly and cost-effectively. Which approach should they use?

Accepted Answer

Use the batch API to process requests asynchronously.. The batch API is designed for high-throughput, asynchronous processing of large volumes of requests, making it ideal for generating synthetic data at scale. It allows the lab to submit many prompts in a single batch, which Azure OpenAI processes efficiently, reducing both cost and time compared to real-time processing.

Answer

Fine-tune a model to generate the data locally.

Answer

Use the streaming API with multiple concurrent connections.

Answer

Deploy a model on Azure Functions and call it in parallel.

Question 16

Which TWO actions should you take to reduce the cost of using Azure OpenAI for a chatbot that handles high traffic?

Accepted Answer

Use the batch API for non-real-time requests.. Option B is correct because the batch API allows you to submit non-real-time requests that are processed asynchronously, often at a lower cost per token compared to real-time inference. This is ideal for chatbot scenarios where some queries (e.g., historical analysis or bulk processing) do not require immediate responses, reducing overall Azure OpenAI consumption costs.

Answer

Increase max_tokens to reduce the number of requests.

Answer

Lower the temperature to 0 to reduce token usage.

Answer

Fine-tune the model to reduce prompt length.

Question 17

A financial services firm wants to use Azure OpenAI to generate investment advice summaries. They must ensure that the model does not produce any advice that could be interpreted as personalized financial advice. What is the most effective strategy?

Accepted Answer

Use a system message that instructs the model to avoid personalized advice and apply strict content filtering.. Option B is correct because Azure OpenAI's system messages allow you to set the model's behavior and constraints at the conversation level, which is the most direct and effective way to enforce a policy like avoiding personalized financial advice. Combined with Azure's content filtering (which can block harmful or restricted content), this approach provides both instruction-based and filter-based guardrails without requiring model retraining or relying solely on example-based prompting.

Answer

Set temperature to 0 and top_p to 0 to make outputs deterministic.

Answer

Provide few-shot examples of disclaimers in the prompt.

Answer

Fine-tune the model on a dataset of generic financial summaries.

Question 18

Which TWO Azure services can be used together with Azure OpenAI to implement a Retrieval-Augmented Generation (RAG) solution?

Accepted Answer

Azure Cosmos DB. Azure Cosmos DB is correct because it can serve as a vector database for storing and querying embeddings generated by Azure OpenAI. When combined with Azure AI Search, which provides hybrid search (vector + keyword) and indexing capabilities, it enables a complete RAG pipeline where relevant documents are retrieved and passed as context to the OpenAI model for grounded generation.

Answer

Azure Blob Storage

Answer

Azure SQL Database

Answer

Azure Functions

Question 19

A developer wants to deploy a custom generative AI model using Azure Machine Learning. Which compute target should they choose for low-latency real-time inference?

Accepted Answer

Azure Kubernetes Service (AKS). Azure Kubernetes Service (AKS) is the correct compute target for low-latency real-time inference because it supports horizontal pod autoscaling, GPU acceleration, and can be configured with a low-latency ingress controller (e.g., NGINX or Azure Application Gateway) to route inference requests directly to model containers. AKS also integrates with Azure Machine Learning's real-time inference endpoint, which uses a gRPC or HTTP-based scoring protocol to achieve sub-100ms response times.

Answer

Local deployment

Answer

Azure Batch

Answer

Azure Functions

Question 20

Match each Azure Cognitive Search skill to its capability.