AI-102 EXAM QUESTIONS 2026 MICROSOFT AZURE AI 102 CERTIFICATION COURSE PART-14

This video is part of a playlist designed to help viewers pass the AI-102 exam. The presenter emphasizes that the playlist covers exact case study questions, with previous students successfully passing the exam after using the material. While some drag-and-drop questions have been converted to a different format for better conceptual understanding, the underlying concepts remain the same. The goal is to ensure viewers grasp the concepts thoroughly, which will enable them to solve 100% of the exam questions once the playlist is complete. Several students, including Harit, Baroni, and Arif, have already passed using this playlist. The video then delves into specific AI-102 exam questions. The first question asks which Azure service to use for measuring public perception on social media using natural language processing. The answer is Azure Language service, as it can perform sentiment analysis and key phrase extraction. Content Moderator is incorrect because its purpose is content moderation, not perception analysis. Computer Vision is also ruled out as social media content is primarily text-based. Form Recognizer (now Document Intelligence) is for extracting data from structured documents. The next question concerns configuring the Content Moderator API to identify aggressive and sexually explicit language. The correct settings are "classify" (to enable the analysis of text for these categories) and "Ocp-Apim-Subscription-Key" (for authentication). "Auto-correct" and "PII" are irrelevant for this specific task. "List ID" is for custom term filtering, not general inappropriate language detection. The third question involves a scenario where users with "reader" access to Azure AI Search are unable to perform index management and querying. The solution is to use an API key to grant access for content operations. While the "contributor" or "owner" roles grant broader permissions, they are considered overprivileged and not the most appropriate solution for search-specific operations, which are primarily controlled by API keys. Service principals are for authenticating applications, not directly for user search operations in this context. The fourth question asks for the appropriate Azure Cognitive Service endpoint for a social media extension that converts text to speech, supporting messages up to 400 characters, offering multiple voice options, and minimizing cost. The correct option is the "TTS (Text to Speech) Cognitive Service" with the "voices/list" endpoint, which allows users to select from available voices cost-effectively. Custom voice and long audio synthesis are not suitable for the given requirements. The fifth question asks what type of value is returned for each category in the classification response of the Text Moderation API. The answer is a confidence score ranging from 0 to 1, where 0 indicates low confidence and 1 indicates high confidence that the text belongs to that category. The sixth question presents a scenario where a container needs to use the latest deployable version of a Language Understanding (LUIS) application. Given version 1.1 has a publish date and is newer than version 1.0, it's the latest deployable version. The required actions in sequence are: exporting the model using the "export for container gzip" option, and then running a container mounting the model file. Running a container with the version set as an environment variable is also part of the process but exporting and mounting are crucial first steps for containerization. The seventh question asks what services should be integrated into a chatbot that needs to support chitchat, a knowledge base, multilingual models, and perform sentiment analysis. The recommended integrations are Q&A Maker (for the knowledge base), Language Understanding (LUIS) for understanding user intent and multilingual support, and Text Analytics for sentiment analysis. Dispatch is not necessary as LUIS can handle routing, and Translator is not needed if LUIS supports multilingual models. The eighth question is a code-related question asking what HTTP method should be used to create a new Azure resource for sentiment analysis and OCR, requiring a single key and endpoint, and consolidated billing. The "PUT" method is the most appropriate for creating a new resource, as it is idempotent (meaning it can be called multiple times with the same result) and will create the resource if it doesn't exist or update it if it does. The blank in the HTTP request should be filled with "PUT". The ninth question asks what should come in the second blank of an HTTP request to create a new resource that meets the requirements of single key/endpoint and consolidated billing for AI services. The answer is "Cognitive Services" (or Azure AI Services), as it consolidates various AI capabilities under a single resource. The tenth question describes successfully running an HTTP request using the secondary subscription key to regenerate a key for an Azure Cognitive Service. The result of this request is that the secondary subscription key was reset. The request specifically targets the secondary key and does not involve primary keys or key vaults. The presenter concludes by encouraging viewers to comment and express their engagement with the playlist, which motivates them to publish more videos.