• Home
• Microsoft Exams
• AI-900 - Microsoft Azure AI Fundamentals

AI-900 Microsoft Practice Test Questions and Exam Dumps

Question No 1:

A company employs a team of customer service agents to provide telephone and email support to customers. The company develops a webchat bot to provide automated answers to common customer queries. 

Which business benefit should the company expect as a result of creating the webchat bot solution?

A. increased sales
B. a reduced workload for the customer service agents
C. improved product reliability

Answer: B

Explanation:

The introduction of a webchat bot primarily serves to automate customer support tasks, allowing the company to handle a larger volume of customer queries without burdening customer service agents. By automating responses to common or repetitive queries, the company can significantly reduce the amount of time its customer service team spends on these tasks, freeing them up to handle more complex or specialized customer issues. This results in a reduced workload for the customer service agents, which is the direct benefit of the webchat bot.

Let’s evaluate the other options:

Option A – Increased sales:
While a webchat bot can improve customer engagement and satisfaction, it does not directly correlate to increased sales. Increased sales are typically the result of more targeted marketing strategies, improved product offerings, or sales team efforts, rather than just the introduction of a support bot. While the bot might indirectly enhance customer experience, it doesn't directly drive sales in the way that a sales-focused strategy might. Therefore, A is not the most accurate answer.

Option C – Improved product reliability:
A webchat bot does not directly affect product reliability. Product reliability is typically influenced by factors like design, quality control, testing, and customer feedback on the product itself. The bot, while improving the efficiency of customer service, does not directly influence the physical or functional reliability of the product. Therefore, C is not the correct answer.

In conclusion, the main business benefit the company should expect from creating the webchat bot solution is B – a reduced workload for the customer service agents, as it allows them to focus on more complex issues while the bot handles repetitive inquiries.

Question No 2:

For a machine learning process, how should you split data for training and evaluation?

A Use features for training and labels for evaluation.
B Randomly split the data into rows for training and rows for evaluation.
C Use labels for training and features for evaluation.
D Randomly split the data into columns for training and columns for evaluation.

Correct Answer: B

Explanation:

In machine learning, the primary goal is to train a model on a portion of data and evaluate its performance on another portion to ensure that the model generalizes well to unseen data. The data typically consists of features (input variables) and labels (target variables). Here's how each option works:

• A Use features for training and labels for evaluation: This option is incorrect. In machine learning, features (the input data) are used to train the model, while the model’s output is compared to the labels (the target values) for evaluation. You would not use labels for evaluation, as labels are the ground truth against which the model's predictions are compared. Therefore, this option does not make sense.

• B Randomly split the data into rows for training and rows for evaluation: This option is the correct approach. The typical way to split data is to randomly divide the dataset into two parts: one for training and the other for evaluation (or testing). Usually, the dataset is split in a way that a large portion (like 70-80%) is used for training, and the remaining portion (like 20-30%) is used for evaluation. Rows refer to individual data points (or samples), and splitting the rows helps ensure the training data and evaluation data are both representative of the overall dataset.

• C Use labels for training and features for evaluation: This is incorrect. Labels (the target variables) are not used for training; they are what the model is trying to predict. The model is trained using the features (the input variables), and after training, you use the features from the evaluation set to see how well the model predicts the corresponding labels. Therefore, this option is not valid.

• D Randomly split the data into columns for training and columns for evaluation: This is also incorrect. Columns represent the features (input data) or labels. Splitting data by columns is not how training and evaluation data are typically separated. You should always split by rows, ensuring that the model has varied data in both training and evaluation sets. Splitting by columns would mix up the relationship between features and labels, which is not useful for model training and evaluation.

In conclusion, the correct method for splitting data is to randomly split the rows into separate sets for training and evaluation, making B the correct answer. This approach ensures that both sets are representative of the data and allows for proper evaluation of the model’s performance.

Question No 3:

You build a machine learning model by using the automated machine learning user interface (UI). You need to ensure that the model meets the Microsoft transparency principle for responsible AI. What should you do?

A. Set Validation type to Auto.
B. Enable Explain best model.
C. Set Primary metric to accuracy.
D. Set Max concurrent iterations to 0.

Answer: B

Explanation:

The Microsoft transparency principle for responsible AI emphasizes the importance of ensuring that AI systems are understandable, interpretable, and transparent to users. This principle requires that AI models are explainable and that their decisions can be easily interpreted, particularly when deployed in real-world scenarios.

Let's evaluate the options based on this principle:

• A. Set Validation type to Auto: Setting the validation type to "Auto" determines how the model will evaluate its performance during training (e.g., cross-validation or holdout validation). While important for model evaluation, this setting does not directly contribute to making the model more transparent or interpretable, nor does it help meet the transparency principle for responsible AI.

• B. Enable Explain best model: Enabling the "Explain best model" feature directly aligns with the transparency principle. It ensures that, once the model is trained, you can generate an explanation of how the best model makes its decisions. This explanation could include insights into feature importance, model behavior, and other interpretability aspects, which are critical for meeting the responsible AI transparency standards. This feature provides clear information about the inner workings of the model, making it more understandable and trustworthy to users.

• C. Set Primary metric to accuracy: While choosing a primary metric such as accuracy helps in model evaluation, it does not impact transparency. Accuracy alone does not help users understand how the model works or why it makes certain decisions. Responsible AI requires more than just selecting a metric; it requires interpretability and clarity in the decision-making process, which accuracy alone does not guarantee.

• D. Set Max concurrent iterations to 0: This option controls how many iterations of the model training process can run concurrently. It is related to performance and resource usage rather than transparency or interpretability. Setting this value does not influence the model's ability to be explainable or meet responsible AI guidelines.

In conclusion, to align with Microsoft's transparency principle for responsible AI, enabling the "Explain best model" option is the most appropriate action. This ensures that the model is explainable, making its decisions clearer and more transparent to stakeholders, which is a key aspect of responsible AI.

Therefore, the correct answer is B.

Question No 4:

You are designing an AI system that empowers everyone, including people who have hearing, visual, and other impairments. This is an example of which Microsoft guiding principle for responsible AI?

A fairness
B inclusiveness
C reliability and safety
D accountability

Correct answer: B

Explanation:

Microsoft’s guiding principles for responsible AI are centered around ensuring that AI is developed and used in a way that benefits everyone and is aligned with ethical standards. These principles include fairness, inclusiveness, reliability and safety, and accountability, among others.

In this case, the AI system is designed to empower people with various impairments, such as hearing, visual, and other disabilities. This focus on making AI accessible to a wide range of users, including those with impairments, directly aligns with the inclusiveness principle.

Inclusiveness in the context of AI means ensuring that AI systems are accessible and beneficial to all individuals, including those from diverse backgrounds and abilities. Designing AI that accommodates people with disabilities exemplifies this principle by making sure that the technology works for everyone, regardless of their physical capabilities.

Let’s break down the other options:

• A Fairness refers to ensuring that AI systems treat all people equitably, without bias or discrimination. While fairness is important in the context of AI, the focus here is more on inclusivity, which directly addresses empowering people with impairments.

• C Reliability and safety focus on ensuring that AI systems are dependable and safe to use. While important, this principle is not directly related to the empowerment of people with impairments.

• D Accountability refers to the idea that those who design and deploy AI systems are responsible for their impact, ensuring transparency and oversight. While this is critical, the focus in this scenario is more on the design of the system to be inclusive.

Therefore, the correct answer is B, inclusiveness, as it directly aligns with designing AI that empowers people with impairments.

Question No 5:

You are working on an AI solution for a customer who wants to classify images into different categories (e.g., animals, vehicles, and people). Which of the following Azure services would you use to build this image classification model?

A. Azure Cognitive Services - Computer Vision
B. Azure Machine Learning
C. Azure Cognitive Services - Face API
D. Azure Bot Services

Answer: A. Azure Cognitive Services - Computer Vision

Explanation:

When it comes to implementing AI for specific tasks, the choice of service is critical. In this case, the objective is to classify images into categories such as animals, vehicles, and people. Let's examine each option and understand why Option A is the best choice.

1. Option A: Azure Cognitive Services - Computer Vision
   Azure’s Computer Vision service is specifically designed for analyzing and understanding images. It provides powerful pre-trained models that can identify and classify objects within images. This service can detect and categorize images, read printed text within images, and even analyze video streams. For tasks like image classification (e.g., recognizing animals, vehicles, and people in images), Computer Vision is the most suitable service because it leverages deep learning models that have already been trained on large datasets, making it efficient and effective for your needs. You can use the Custom Vision tool within Computer Vision to build a custom image classification model based on your specific categories (such as animals, vehicles, etc.), enabling high-accuracy results.

2. Option B: Azure Machine Learning
   While Azure Machine Learning is a powerful service for building custom machine learning models, including image classification models, it typically requires more manual setup. This would involve creating a model from scratch or fine-tuning an existing one. While this is a valid approach for experienced data scientists, Computer Vision is a more accessible and specialized service for users who need to quickly deploy AI solutions for image classification without requiring deep ML expertise.

3. Option C: Azure Cognitive Services - Face API
   The Face API is specifically designed for detecting and recognizing human faces in images. It’s a highly specialized tool for facial recognition tasks, not for general image classification. If the customer’s goal is to categorize images into broader categories like animals or vehicles, this service wouldn’t be appropriate.

4. Option D: Azure Bot Services
   Azure Bot Services is used to build conversational agents (chatbots) and does not specialize in image classification tasks. It’s great for scenarios that involve natural language processing (NLP) or automated conversation systems, but it’s not suited for tasks like recognizing or categorizing images.

In summary, Azure Cognitive Services - Computer Vision (Option A) is the most suitable service for building an image classification model in this scenario. It is designed to handle image analysis tasks, is easy to implement, and leverages pre-trained models that can be customized to fit specific categories, making it ideal for quickly deploying AI solutions for image recognition.

Question No 6:

Which of the following Azure services is primarily used for creating and deploying machine learning models with minimal coding?

A. Azure Cognitive Services
B. Azure Machine Learning
C. Azure Bot Service
D. Azure Functions

Answer: B. Azure Machine Learning

Explanation:

In the AI-900: Microsoft Azure AI Fundamentals exam, one of the key areas tested is the understanding of the various AI-related services available on Azure. Let's break down the options and understand why Option B: Azure Machine Learning is the correct choice.

1. Azure Cognitive Services (Option A)
   Azure Cognitive Services is a suite of pre-built APIs that allow developers to add AI capabilities to applications without the need for deep AI or machine learning knowledge. These services are used for specific tasks like computer vision, speech recognition, language understanding, and more. While Cognitive Services is an important part of Azure's AI offering, it is not designed for creating and training custom machine learning models. Instead, it provides ready-to-use AI models for various use cases.

2. Azure Machine Learning (Option B)
   Azure Machine Learning is a cloud-based service that provides a comprehensive environment for building, training, and deploying machine learning models. It supports various development frameworks and languages, including Python and R, and integrates with tools like Jupyter notebooks. Importantly, Azure Machine Learning offers a no-code/low-code interface called Azure Machine Learning Studio, which is designed for users who may not have extensive programming skills but still want to create machine learning models. This service is ideal for creating and deploying machine learning models with minimal coding, making it the correct answer for this question.

3. Azure Bot Service (Option C)
   The Azure Bot Service is designed for building intelligent bots that can interact with users through conversational interfaces, such as chatbots or virtual assistants. While bots can integrate with AI and machine learning capabilities (often through Cognitive Services), the Bot Service itself is not used to create machine learning models. Its primary purpose is to facilitate bot creation, rather than model development.

4. Azure Functions (Option D)
   Azure Functions is a serverless compute service that allows you to run code in response to events or triggers, such as HTTP requests, changes in data, or messages from other services. While it can be used in the context of machine learning (for example, triggering machine learning inference), it is not a service specifically designed for developing or deploying machine learning models.

Key Takeaways:

• Azure Machine Learning is the go-to service for creating and deploying custom machine learning models, especially if you're looking for tools to help with minimal coding.

• Azure Cognitive Services provides ready-to-use AI services for specific tasks like speech, vision, and language but is not intended for building custom machine learning models.

• Azure Bot Service focuses on creating intelligent bots, while Azure Functions provides serverless compute but does not specialize in AI or machine learning.

For those studying for the AI-900 exam, it’s important to familiarize yourself with the key Azure services that enable AI and machine learning tasks. Azure Machine Learning stands out as the most comprehensive solution for machine learning model development, especially for users who wish to minimize coding.

Question No 7:

Which of the following is a key benefit of using Azure Cognitive Services for AI applications?

A. It allows you to build custom AI models without any coding experience.
B. It requires you to manually manage infrastructure for running AI models.
C. It offers pre-built AI models for common tasks like speech recognition, sentiment analysis, and computer vision.
D. It provides tools for developing only speech-related AI models.

Answer: C. It offers pre-built AI models for common tasks like speech recognition, sentiment analysis, and computer vision.

Explanation:

Azure Cognitive Services is a collection of pre-built APIs that allow developers to easily integrate AI capabilities into applications without needing deep expertise in AI or machine learning. This includes tasks like computer vision, speech recognition, sentiment analysis, and language translation.

1. Option C: Pre-built AI models
   The key benefit of using Azure Cognitive Services is that it provides pre-built AI models for various tasks such as speech recognition, sentiment analysis, image recognition, and more. These APIs allow developers to integrate powerful AI functionalities into their applications quickly and with minimal coding required.

2. Option A: No coding experience required
   While Cognitive Services minimizes the need for deep AI knowledge, some basic coding is still required to call the APIs and integrate them into an application. It's not completely "no-code."

3. Option B: Managing infrastructure
   Azure Cognitive Services is a managed service, meaning Microsoft takes care of the infrastructure for you. There's no need for developers to manually manage servers or AI model hosting.

4. Option D: Speech-related AI models only
   Although Azure Cognitive Services offers powerful tools for speech, it is not limited to just speech-related AI. It also supports other AI models for computer vision, language understanding, and more.

The main advantage of Azure Cognitive Services is that it offers a set of pre-built AI models that help solve common tasks such as sentiment analysis, image classification, and speech recognition, all with minimal setup and coding.

Question No 8:

Which of the following Azure services can be used to build, train, and deploy custom machine learning models, especially for more complex or specialized AI workloads?

A. Azure Cognitive Services
B. Azure Bot Services
C. Azure Machine Learning
D. Azure Speech Services

Answer: C. Azure Machine Learning

Explanation:

Azure Machine Learning is a comprehensive platform that enables users to build, train, and deploy custom machine learning models. It is ideal for more complex or specialized AI workloads that require custom models or training on large datasets.

1. Option C: Azure Machine Learning
   Azure Machine Learning is a robust cloud-based environment for developing, training, and deploying custom machine learning models. This platform provides tools for data scientists and developers to create custom models, experiment with algorithms, manage training datasets, and deploy models into production environments.

2. Option A: Azure Cognitive Services
   Azure Cognitive Services provides pre-built, ready-to-use AI capabilities like sentiment analysis, language translation, and computer vision. However, it is not designed for building custom machine learning models. It’s more about applying AI models to your data, rather than creating new models.

3. Option B: Azure Bot Services
   Azure Bot Services is used for building conversational bots, including chatbots. It is not intended for training or deploying machine learning models and is focused on bot development rather than custom AI workloads.

4. Option D: Azure Speech Services
   Azure Speech Services offers capabilities for speech-to-text, text-to-speech, and speech translation. It focuses on processing and converting speech, but it is not a platform for building custom machine learning models.

Azure Machine Learning is the best service for building, training, and deploying custom machine learning models, especially when working with more complex or specialized AI applications. It provides a comprehensive suite of tools for data scientists and developers to manage the full ML lifecycle.

These two questions address key Azure AI services, focusing on Azure Cognitive Services for pre-built AI models and Azure Machine Learning for building custom models. Both are essential topics for preparing for the AI-900 Microsoft Azure AI Fundamentals exam.

Question No 9:

You need to build a solution that can automatically translate text from one language to another in real-time within your application. Which Azure service should you use?

A. Azure Cognitive Services - Text Analytics
B. Azure Cognitive Services - Translator
C. Azure Cognitive Services - Custom Vision
D. Azure Machine Learning

Answer: B. Azure Cognitive Services - Translator

Explanation:

The Azure Cognitive Services - Translator is designed specifically for translating text between different languages in real-time.

1. Option B: Azure Cognitive Services - Translator
   The Translator service is part of Azure Cognitive Services and is built to handle text translation tasks. It can translate text between multiple languages in real-time and provides capabilities such as language detection and batch translation. This is the perfect service for any application that needs real-time translation of user input, customer interactions, or content.

2. Option A: Azure Cognitive Services - Text Analytics
   The Text Analytics service is used for tasks like sentiment analysis, language detection, and key phrase extraction. While it includes language detection, it does not offer translation capabilities, making it unsuitable for real-time text translation.

3. Option C: Azure Cognitive Services - Custom Vision
   Custom Vision is designed for image classification and object detection. It’s not used for text analysis or translation, so it does not apply to this scenario.

4. Option D: Azure Machine Learning
   Azure Machine Learning is a powerful tool for building custom models for machine learning tasks. However, it doesn’t provide out-of-the-box solutions for text translation. You could use Azure Machine Learning to build a custom translation model, but Translator is a simpler, pre-built service for real-time translation.

For automatic real-time text translation, Azure Cognitive Services - Translator is the best choice. It’s designed specifically for this purpose, providing robust translation features.

Question No 10:

You need to build a custom image classification model that can recognize and label various types of vehicles in images, such as cars, trucks, and motorcycles. Which Azure service should you use?

A. Azure Cognitive Services - Custom Vision
B. Azure Machine Learning
C. Azure Cognitive Services - Face API
D. Azure Cognitive Services - Speech API

Answer: A. Azure Cognitive Services - Custom Vision

Explanation:

To build a custom image classification model for recognizing different types of vehicles, Azure Cognitive Services - Custom Vision is the ideal service.

1. Option A: Azure Cognitive Services - Custom Vision
   Custom Vision allows you to build custom image classification models with ease. By uploading and labeling images of different vehicles (cars, trucks, motorcycles, etc.), you can train a model to classify images of vehicles accurately. This service is specifically designed for users who want to create custom models for image classification tasks without requiring deep machine learning expertise.

2. Option B: Azure Machine Learning
   While Azure Machine Learning can be used to build a custom image classification model, it is more complex and requires more effort, as it involves creating and training models manually. Custom Vision is a simpler and more specialized solution for tasks like vehicle classification.

3. Option C: Azure Cognitive Services - Face API
   The Face API is specialized for face detection and recognition. It is not designed for general object or vehicle classification tasks.

4. Option D: Azure Cognitive Services - Speech API
   Speech API is used for tasks related to speech recognition, such as converting spoken language into text. It does not deal with image classification or object detection.

For custom image classification tasks, Azure Cognitive Services - Custom Vision is the best choice. It simplifies the process of training a model to recognize and label objects, such as vehicles, in images.