AI-102 Exam Dumps - Designing and Implementing a Microsoft Azure AI Solution

For a social media messaging app that needs to determine the language of short text in real time, the most appropriate service is Azure AI Translator. Translator provides a low-latency language detection operation for text, returning the detected language code and a confidence score, and it’s optimized for interactive scenarios like chat and messaging. You can use detection on its own or feed the result directly into translation if needed.

Why not the others:

Azure AI Speech is for audio (speech-to-text, text-to-speech, speech translation), not text-only language detection.

Azure AI Language has language detection within its Text Analytics features, but Translator’s detect endpoint is the recommended, streamlined choice for real-time translation pipelines and chat apps.

Azure AI Foundry Content Safety is for moderation (toxicity, harassment, hate), not language identification.

Microsoft References

Azure AI Translator — Text Translation overview (includes language detection).https://learn.microsoft.com/azure/ai-services/translator/

Text Translation REST API v3.0 — Detect operation.https://learn.microsoft.com/azure/ai-services/translator/reference/v3-0-detect

Quickstart: Use the Translator text client libraries (shows detect + translate flow).https://learn.microsoft.com/azure/ai-services/translator/quickstart-text-sdk

Azure Video Indexer can ingest a video by URL as long as the URL is publicly accessible and directly downloadable (no authentication or HTML landing page). Creating a download link in OneDrive produces a direct-file URL that Video Indexer can fetch and index without you re-uploading the 20-GB file through your browser. A generic OneDrive “sharing link” often lands on a web page (not a direct file) and may require authentication, which Video Indexer cannot crawl. Uploading locally is possible but unnecessary and time-consuming for a 20-GB video.

Microsoft References

Upload/index videos in Video Indexer (supports uploading from a public URL).https://learn.microsoft.com/azure/azure-video-indexer/upload-index-videos

Video Indexer quotas and limits (file size limits; uploading by URL).

1. Number of Projection GroupsA projection group in a knowledge store is an array of projections (tables, objects, or files) that are stored together.

In the provided JSON, the projections array contains the following two top-level sections:

The array of tables:

JSON

"tables": [

{ ... },

{ ... }

]

The array of objects:

JSON

"objects": [

{ ... }

]

Each of these top-level arrays (tables and objects) represents a distinct projection group in the knowledge store. Therefore, there are two projection groups.

2. Image SavingThe files projection is used to save images, normalized images, or application files generated during the enrichment process.

In the provided JSON, the files array is present but is empty:

JSON

"files": []

Since there are no file projections defined, the images (or any files) generated during the enrichment process will not be saved to the knowledge store. They are processed in the pipeline but not persisted in the output storage defined by the knowledgeStore configuration.

This question aligns with Microsoft Responsible AI principles (Fairness, Reliability & Safety, Privacy & Security, Inclusiveness, Transparency, and Accountability).

The scenario:

Sentiment Analysis is used to determine staff bonuses (a financial impact).

Automated sentiment classification can be biased or produce incorrect results.

Decisions that affect people’s financial or legal status should never be made fully automated.

Analysis of the options:

A. Add a human review and approval step → Correct. This ensures accountability and fairness, reducing harm from potential model bias or errors.

B. Include results even with low confidence → Wrong. That increases risk of incorrect decisions.

C. Use all surveys including deleted accounts → Wrong. Violates privacy and compliance.

D. Publish raw survey data → Wrong. Violates privacy and security principles.

Therefore, the best responsible AI-compliant action is to add human review before making financial decisions.

Correct Answer for Q186: A

Microsoft Responsible AI principles

Responsible use of AI in decision-making

Microsoft Reference

Explanation:

By default, Azure AI Language may log request and response data for monitoring.

Setting the loggingOptOut parameter to true ensures that input data is not persisted after analysis.

piiCategories specifies which PII types to detect.

showStats enables request/response statistics.

model-version specifies which model version to use.

Correct Answer: A

To “recognize geographic locations” in Azure AI Search enrichment, add the EntityRecognitionSkill to your skillset. This built-in cognitive skill performs named entity recognition (NER) and extracts entities such as locations, people, and organizations from text.

EntityLinkingSkill links already-recognized entities to a knowledge base (e.g., Wikipedia) and is not required just to detect locations.

DocumentExtractionSkill parses file content/structure.

AzureOpenAIEmbeddingSkill creates vector embeddings, not entities.

Microsoft References

Azure AI Search – Entity recognition skill (extracts entities including locations).

Azure AI Search – Entity linking skill (disambiguates entities to a knowledge base; not for initial recognition).

The percentage of false positives is [0].

The value for the number of true positives divided by the total number of true positives and false negatives is [100].

The screenshot shows the Performance tab of a Custom Vision object detection model after training.

Two key performance metrics are in focus:

False Positives (FP):

False positives occur when the model incorrectly predicts an object that isn’t present.

In the provided exhibit, the model shows 0% false positives.

This means it did not incorrectly classify any non-object as an object.

Recall (True Positives ÷ (True Positives + False Negatives)):

This measures how many of the actual objects were correctly detected.

Formula: Recall = TP / (TP + FN).

In the performance report, recall is 100%, meaning the model correctly detected all the actual objects (no false negatives).

False positives: 0%

Recall (TP ÷ (TP + FN)): 100%

Evaluating the answer choices:

Correct Answers:

False positives → 0

Recall → 100

Custom Vision performance metrics

Precision, recall, and accuracy in machine learning

Microsoft References