Microsoft Updated AI-103 Exam Questions and Answers by nolan

The correct observability signal is B. groundedness evaluation metrics . In a RAG solution, the key diagnostic question is whether the generated answer is supported by the retrieved context. Microsoft Foundry’s built-in evaluator reference defines Groundedness as the metric that measures how grounded the response is in the retrieved context, with scoring that indicates whether the model’s claims are supported by the provided source material.

This matches the issue after a content update. If retrieved chunks are stale, misleading, incomplete, or poorly aligned with the user query, groundedness results can show that generated responses are not reliably supported by the retrieved documentation. The RAG evaluator guidance explains that groundedness focuses on whether the response avoids content outside the grounding context, while other process metrics such as retrieval evaluate how relevant the retrieved chunks are. Latency traces are useful for performance troubleshooting, not response accuracy. Indexer status can reveal ingestion failures, but it does not show whether retrieved content is influencing generated answers negatively. Prediction drift is a model monitoring concept and is not the primary signal for RAG grounding quality. Reference topics: Microsoft Foundry observability, RAG evaluators, groundedness, retrieved context, and response quality evaluation.

The correct answer is C. Split the mixed-language segments into single-language segments and translate each segment separately . Azure Translator has a known limitation for text that contains mixed-language input within the same sentence or segment. Microsoft’s Translator known-issues guidance states that the Text translation API does not support translating sentences that contain mixed-language text and that translations can be incorrect or incomplete when a single sentence includes multiple languages. The documented workaround is to specify the intended source language, remove the mixed-language sentence, or split the text into single-language segments.

In this scenario, specifying English for all segments would fail when Spanish content is present. Automatic language detection can detect a dominant language, but it does not reliably solve a segment that contains both English and Spanish. Document translation is also not the right fix because the issue is mixed-language content inside transcript segments, and Microsoft notes similar incomplete results for mixed-source-language documents. Splitting the transcript into language-homogeneous segments ensures each translation request has a clear source language and lets the entire transcript be translated reliably. Reference topics: Azure Translator in Foundry Tools, text translation, language detection, mixed-language input, and transcript preprocessing.

Use GPT-4 because suggesting product names from a textual description is a natural-language generation task. The application can supply the product’s characteristics, intended audience, tone, and naming constraints in a prompt, and GPT-4 can generate candidate names that reflect those requirements. GPT models interpret natural-language instructions and generate corresponding textual completions, making them appropriate for ideation, summarization, content drafting, and conversational applications.

The other model categories serve different modalities. DALL-E is an image-generation model that creates visual content from natural-language prompts; it is not intended primarily to produce product-name lists. Whisper is a speech-recognition model used to transcribe audio into text. Embedding models convert text into numerical vectors for similarity search, clustering, classification, and retrieval scenarios; they do not generate human-readable product names. Microsoft’s Foundry model documentation distinguishes chat and generative models from image-generation, audio, and embedding model families.

The app should therefore deploy GPT-4, construct a prompt containing the product description and naming criteria, and consume the generated textual response through the model inference endpoint.

Study Guide alignment: Select and deploy appropriate generative models, integrate deployed models into applications, and implement prompt-driven content-generation workflows.

Comprehensive and Detailed 150 to 250 words of Explanation From Azure AI Apps and Agents Developer/Course Guide/topics:

Spatial Analysis is the appropriate capability because the requirement is to analyze a video stream and determine how many people are present in a defined physical space.

Spatial Analysis provides pretrained person-detection and spatial-event functionality that can monitor occupancy, presence, and movement without requiring the developer to build a custom person-detection model. The application can use those detections to determine whether more than one person appears within the monitored area and generate an appropriate violation event.

Custom Vision object detection could theoretically be trained to detect people, but doing so would require data collection, labeling, model training, deployment, and additional application logic. That conflicts with the requirement to minimize development effort.

Speech-to-text analyzes microphone audio and cannot determine how many people are visually present. Azure AI Video Indexer derives insights from video but is not the specialized real-time physical-space monitoring solution represented by this scenario.

The AI-103 Study Guide includes implementing solutions that identify objects, components, or regions within images or video and building video understanding workflows, making Spatial Analysis the appropriate pretrained service choice. Microsoft Learn

====================