Microsoft Azure AI Fundamentals Practice Test (AI-900)

They should use Computer Vision. This workload enables machines to interpret and understand visual information from images or videos. In this scenario, the retail company aims to analyze images from cameras to identify and categorize products, which is a typical application of Computer Vision.

Natural Language Processing deals with understanding and processing human language, which is not applicable here.

Knowledge Mining involves extracting information from large volumes of text but doesn't directly address image analysis.

Generative AI focuses on creating new content, such as text or images, and is not suited for product identification tasks.

Document intelligence workloads involve extracting structured information from unstructured or semi-structured documents like invoices, receipts, and forms. A service that extracts key information from invoices and receipts is an example of this, as it automates data extraction from documents.

The other options represent different AI workloads: translating spoken language into text is speech recognition (natural language processing), recommending products based on user preferences is personalization, and identifying objects in images is computer vision.

To evaluate the model's performance on unseen data, helping to prevent overfitting - This is the correct answer. Dividing a dataset into training and validation subsets allows the model to learn from the training data while being evaluated on a separate validation set. This helps assess the model's ability to generalize to new, unseen data, preventing overfitting where the model memorizes the training data but performs poorly on new data.

To increase the total amount of data available for training - This is incorrect because splitting the data reduces the amount of data available for training. The goal is to use a portion for training and another for validation, not to increase the total amount of data.

To reduce the training time by using smaller datasets - Splitting the data does not specifically aim to reduce training time. It’s more about evaluating model performance and preventing overfitting.

To ensure the model memorizes the training data perfectly - The goal is not for the model to memorize the training data. In fact, memorization (overfitting) is something practitioners want to avoid. The focus is on ensuring the model generalizes well to new, unseen data.

Content moderation workloads in AI are designed to automatically detect and filter inappropriate or harmful content, such as offensive language, violent images, or spam. This helps maintain a safe and welcoming environment for users.

Language Modeling is the appropriate AI capability for this scenario because it can generate and predict text based on context, allowing the application to suggest probable sentence completions.

Sentiment analysis determines the emotional tone of text, entity recognition identifies and classifies entities within text, and key phrase extraction identifies the main topics or themes. These capabilities do not generate text predictions.

Facial analysis features of the Azure AI Face service detect faces in an image and can predict certain attributes-such as an estimated age-for each detected face (this attribute is available only to approved customers with limited access). Object detection locates generic objects but does not return facial attributes. Text Analytics works with text data, and Speech Recognition works with audio; neither service can estimate age from images.

Generative AI models learn from massive text and image corpora that contain historical and social biases. Those biases can surface or even amplify in model outputs. Responsible AI guidance from Microsoft requires developers to continuously identify, measure, and mitigate potential harms by combining automated tools (such as content filters) with human-in-the-loop reviews, red-team testing, and ongoing monitoring. Ignoring these steps risks unfair, stereotyped, or otherwise harmful generations. Therefore, the only correct statement is that bias can occur and must be addressed through sustained human oversight.

Classification - This is the correct answer. Fraud detection is a classification task because the model needs to classify each transaction as either fraudulent or legitimate, which involves assigning data points to predefined categories or labels.

Regression is used for predicting continuous numerical values for example sales forecasts or prices, not for classifying transactions into categories such as "fraudulent" or "legitimate."

Clustering is an unsupervised learning technique used to group data based on similarities, but it is not suitable for determining whether a transaction is fraudulent or legitimate, which requires labeled data and a classification approach.

Dimensionality Reduction is used to reduce the number of features in the data, typically for improving performance or visualization, but it is not a task in itself for determining fraud or legitimacy.

Using language generation models to produce creative content - This is the correct answer. Language generation models, such as GPT, are ideal for automatically creating creative content like slogans and taglines by generating text based on input prompts or the product's features.

Applying data analysis and visualization techniques - This method is used to analyze and present data in a visual format

Implementing product recommendation algorithms - Product recommendation algorithms are used to suggest products to customers based on their behavior or preferences

Utilizing sentiment analysis for text assessment - Sentiment analysis is used to assess the emotional tone of text

Content Synthesis tasks - This is the correct answer. Content synthesis tasks involve generating new content that is similar to the examples the model has been trained on. This is typically seen in generative models, where the system can create new, original content such as text, images, or music based on the patterns it has learned.

Anomaly Detection focuses on identifying outliers or unusual patterns in data, not on generating new content. It is used for detecting abnormal behavior or data points.

Data Classification involves categorizing data into predefined classes or categories, such as labeling emails as spam or not. It does not involve generating new content.

Regression Analysis is used to predict numerical values based on historical data, such as forecasting sales or stock prices. It is not focused on generating new content.

Allowing stakeholders to audit and review the system - This is the correct answer. Allowing stakeholders to audit and review the AI system ensures transparency and accountability in the development process. It enables oversight, helps identify potential issues, and ensures that the system meets ethical and legal standards.

Maintaining complete confidentiality of development processes - While confidentiality is important for protecting intellectual property, it can hinder accountability by limiting transparency and oversight from stakeholders who need to ensure the system is ethical and fair.

Automating decision-making without human oversight - Automating decision-making without human oversight can reduce accountability, as there would be no mechanism for checking or correcting decisions made by the AI system, especially in critical areas like healthcare or finance.

Prioritizing only performance metrics during development - Focusing solely on performance metrics can lead to overlooking ethical concerns and accountability issues. A holistic approach that considers fairness, transparency, and accountability is essential in AI development.

The label in a machine learning dataset is the variable that you aim to predict. In this scenario, the company wants to forecast future sales, so total sales is the label.

The other columns are features used to make the prediction.

Deep learning techniques have the capability to automatically learn complex patterns and features from raw data like images. This ability eliminates the need for manual feature engineering, allowing the model to identify intricate patterns within the data that may be difficult to extract manually.

The other options are incorrect because deep learning models typically require substantial computational resources and large datasets to perform effectively. Additionally, they do not depend on manual feature extraction methods.

Computer Vision is the appropriate AI workload for analyzing visual data to identify and track objects or patterns within images or video. In this scenario, the team needs to process visual data to recognize different animal species, which is a typical application of Computer Vision.

Natural Language Processing (NLP) deals with understanding and generating human language, which is not relevant here.

Knowledge Mining involves extracting information from large datasets but doesn't specifically handle visual recognition tasks.

Document Intelligence focuses on extracting information from documents, which does not apply to analyzing images or videos.

Analyzing images and extracting visual information - This is the correct answer. The Azure AI Vision service provides developers with capabilities to analyze images and extract visual information. This includes tasks such as image classification, object detection, and optical character recognition (OCR), helping developers to gain insights from images.

Transcribing spoken language into text - This is the functionality of speech recognition services, not the Azure AI Vision service. Speech recognition is focused on converting spoken language into written text.

Translating text between different languages - This is the functionality of Azure Translator service, which is designed for translating text between languages, not related to image analysis.

Performing sentiment analysis on text data - This is the functionality of text analytics services, which are used for tasks like sentiment analysis, key phrase extraction, and language detection, but it is not related to image analysis.

This statement is False.

Technology that converts spoken language into written text is known as speech recognition. Text generation involves creating new text content, often based on patterns learned from existing data, but does not specifically refer to transcribing spoken words.

Use the service's facial attribute analysis capability (invoking the Detect API with the returnFaceAttributes parameter). In addition to the face rectangle, the call can return headPose, blur, mask, accessories, qualityForRecognition, occlusion and other attributes for every detected face, enabling per-user customization.

Face Detection without attributes supplies only bounding-box coordinates (and optionally a faceId).

Face Identification compares a detected face to a person group to find who it is.

Face Similarity Matching (Find Similar) returns faces that visually resemble a given face but does not expose individual attributes.

Image Classification - This is the correct answer. Image classification is the most appropriate solution for assigning a single label to each image based on the main object or feature present in the image. It categorizes the entire image into predefined classes for example "dog," "car," "cat" without identifying specific objects' locations within the image.

Object Detection - Object detection is used for identifying and locating multiple objects within an image, often with bounding boxes. While it can classify objects, its primary focus is on locating objects, making it more complex than image classification for this task.

Optical Character Recognition (OCR) - OCR is specifically used for extracting text from images or documents, not for classifying images based on their content.

Facial Detection - Facial detection focuses on detecting human faces within images. It is not used for general image classification based on the main object in the image.

Predicting the content category of an image - This is the correct answer. Image classification involves assigning a label or category to an image based on its content. For example, an image classification solution might categorize an image as "cat," "dog," or "car" based on the objects it contains.

Identifying and locating objects within an image - This describes object detection, not image classification. Object detection involves both identifying objects and locating them within the image (often with bounding boxes), while classification only assigns a category label.

Extracting text from images for text analysis - This is Optical Character Recognition (OCR), which is focused on extracting text from images, not classifying the entire image into categories.

Detecting facial features to analyze emotions - This is facial recognition or emotion detection, which focuses on detecting and analyzing facial features, often for applications like identifying emotions or individuals, but it is not image classification.

Azure OpenAI Service offers advanced language models that can generate human-like text based on input prompts, making it suitable for content creation and code suggestions. The other services listed do not provide text generation capabilities.

Azure Translator Service is designed for translating text between languages.

Azure Cognitive Search enables indexing and querying of content.

Azure Text Analytics provides features like sentiment analysis and key phrase extraction but does not generate new text.