AB-410 Exam Questions

Tokenization is the process of converting text into a format (tokens) that the transformer architecture can understand and manipulate.

The Vector Store holds the numerical representations (embeddings) of your data, allowing the system to quickly find relevant context for a prompt.

The “Adult“ feature in Azure AI Vision‘s Image Analysis is specifically tuned to detect mature, racy, or violent content.

The Fairness principle mandates that AI systems should treat all people equally and avoid biased outcomes.

Power Automate provides the connectivity to external messaging services to handle transactional tasks like sending SMS codes.

Cosine similarity is a metric used to determine how similar two vectors are by measuring the cosine of the angle between them.

Groundedness evaluates whether the model’s response is factually derived from the reference context provided, rather than fabricated.

Language Detection identifies the source language, which can then be used to route the conversation or translate the output.

Azure AI Document Intelligence is specifically designed to extract text, key-value pairs, and structural data from documents like invoices.

Fine-tuning retrains the model on specific data to change its behavior or knowledge; RAG provides external context to a pre-trained model at query time.

PII Redaction involves identifying and removing or masking sensitive data like names and IDs before the data is logged or processed.

Frequency Penalty penalizes tokens based on their existing frequency in the text, discouraging repetitive output.

The Text Summarization pre-built model is designed to provide concise versions of long-form text documents like handbooks.

Dataverse supports geospatial data types and functions that allow apps to calculate distances and find records based on geographic coordinates.

Microsoft is introducing a native Vector data type in Dataverse to support efficient storage and retrieval of embeddings for RAG-based applications.

The System Prompt acts as the “inner instructions“ that tell the AI how to behave (e.g., “You are a helpful assistant that only uses professional language“).

Azure AI Vision is needed to extract the text from the image (OCR), and Azure AI Translator is used to convert that text to the target language.

Semantic Kernel is an open-source SDK that lets you easily combine AI services with conventional programming languages to build agents.

Text Recognition (OCR) extracts the text, Translator changes the language, and Text-to-Speech (TTS) provides the audio output.

Copilot Studio allows you to upload files (including Excel or PDFs) directly as Knowledge Sources to enable the bot to answer questions based on that data.

Azure AI Content Safety is a dedicated service that checks both input (prompts) and output (completions) for harmful content categories.

Vector search uses embeddings to find results that are conceptually related to the query, even if the exact words don‘t match.

Grounding provides the AI with the specific “truth“ (like a product list) to ensure its generated answer is factually accurate.

Security Roles and Business Units define the scope of data access, ensuring AI agents respect organizational boundaries and permissions.

Azure AI Studio is the unified platform for building, evaluating, and deploying AI models, including the Model Catalog for open-source options.

A prompt is the text or instruction given to an LLM that guides what it should generate or do.

Object Detection can be trained to identify and locate specific items (like logos or products) within an image.

Copilot Studio is the orchestration layer used to build the “brains“ (agents) that can then be hosted within Power Apps or other channels.

SLMs (like the Phi-3 family) have fewer parameters, making them faster and cheaper to run on edge devices compared to massive LLMs.

Coherence measures the logical consistency and smooth transition of ideas within the AI-generated text.

The Prediction model analyzes historical patterns to estimate the likelihood of future outcomes (Yes/No or binary results).

The On-premises Data Gateway acts as a bridge, providing quick and secure data transfer between on-premises data and Microsoft Cloud services.

The Copilot pane in the Power Pages Design Studio allows users to describe the form they need, and it automatically creates the Dataverse table and form layout.

The Power Platform Admin Center is where administrators configure DLP policies to control how data is shared through connectors and AI agents.

The “Filterable“ attribute allows the field to be used in $filter queries to narrow down search results based on specific criteria.

Generative AI models excel at understanding long-form text and condensing it into concise summaries while maintaining key facts.

The Sentiment Analysis feature is a core capability of Azure AI Language, used to detect positive, negative, or neutral tones in text.

Adaptive Cards are a platform-agnostic way to show snippets of UI (like buttons, images, and text blocks) generated by an AI bot.

Inclusiveness ensures that AI applications empower everyone and address the needs of diverse populations, regardless of ability or background.

Custom Translator allows you to train the translation engine on your own domain-specific terminology (like legal or medical jargon) for better accuracy.

Reducing Max Tokens sets a lower limit on the output length, forcing the model to be more brief.

AI Builder provides a user-friendly, no-code interface within Power Apps and Power Automate to train and use custom document processing models.

Input variables allow the prompt to be dynamic, pulling in data from Dataverse or a Power Automate flow to provide specific context.

Dataverse Plug-ins are custom code snippets that execute on the server side to handle complex data validation or external integrations.

Grounding ensures the model‘s responses are based strictly on the provided context (the legal documents) and prevents hallucinations.

While many settings exist, ensuring the bot can handle rich media like Adaptive Cards is essential for interactive intelligent applications.

The Indexer is the crawler that connects to a data source, reads the data, and serializes it into the search index.

Azure AI Document Intelligence includes advanced OCR capabilities specifically tuned for both printed and handwritten text in various document formats.

Prompt Flow in Azure AI Studio allows for the creation, evaluation, and iterative testing of complex prompt logic and AI workflows.

Models like GPT-4o-mini or GPT-3.5 Turbo are optimized for speed and cost-efficiency compared to larger models like GPT-4.

The “max_tokens“ parameter sets a hard limit on the length of the generated response (though tokens do not map perfectly to word counts).

A Confusion Matrix is a table used to describe the performance of a classification model on a set of data for which the true values are known.

RAG allows the model to retrieve the most current information from an external data source (the 2026 catalog) before generating a response.

Azure AI Content Safety provides built-in filters to detect and block hate, violence, self-harm, and sexual content in AI-generated text and images.

Azure AI Studio provides evaluation tools to calculate metrics like groundedness, relevance, and coherence against a test dataset.

Reliability and Safety focus on ensuring the AI system works as intended and does not cause harm during failures or edge cases.

Azure AI Speech provides Speech-to-Text and Speech Translation capabilities that can handle multi-language translation in real-time.

To authenticate and send requests, an application needs the specific endpoint (API Base URL) and a valid API Key (or Token).

The System Message instructs the model on how to behave, what tone to take, and what constraints to follow during the session.

Agentic Workflow involves the AI agent making logical decisions and calling external actions (Plugins/Flows) to complete a complex business process.