AI-900 Azure AI Fundamentals Sample Questions & Practice Tests

Get exam-ready with these updated AI-900 sample questions for Azure AI Fundamentals certification. Tailored for beginners, this resource covers essential AI concepts, responsible AI, machine learning models, computer vision, and natural language processing—all mapped to Microsoft’s AI-900 exam domains. These sample questions simulate real exam scenarios to test your knowledge and readiness. Whether you're preparing with AI-900 dumps, practice tests, or full-length mock exams, this guide boosts your confidence and comprehension for Microsoft Azure AI Fundamentals certification success in 2025.

AI-900 Azure AI Fundamentals Sample Questions List :

1. Extracting key insights from structured and unstructured data sources is a feature of what type of workload?

1. Natural Language Processing

2. Computer Vision

3. Anomaly detection

4. Knowledge Mining

2. Training of a custom image classification model“ can be performed by using the Computer Vision service.

1. No

2. Yes

3. Your team member is planning to use the Form Recognizer pre-built receipt model.Which kind of Azure resource should you recommend to create?

1. Computer Vision

2. Form Recognizer

3. Face Service

4. A banking system, that predicts whether a loan will be repaid or not, is an example of what type of machine learning?Select the correct option.

1. Classification

2. Clustering

3. Regression

5. Identifying suspicious sign-ins by looking for deviations from normal patterns is an example of anomaly detection.

1. Yes

2. No

6. Which type of machine learning should you use to predict the sale price of auctioned items.

1. Regression

2. Clustering

3. Classification

7. A media company is implementing an AI system that entitles everyone including people with disabilities such as vision impairment, deaf or hard of hearing. Identify the Microsoft guiding principle for responsible AI which the company is trying to implement.

1. Inclusiveness

2. Accountability

3. Reliability and safety

4. Fairness

8. An automobile dealership wants to use historic car sales data to train a machine learning model. The model should predict the price of a pre-owned car based on its make, model, engine size, and mileage. What kind of machine learning model should the dealership use automated machine learning to create?

1. Classification

2. Time Series

3. Regression

9. An automated chat to answer questions about refunds and exchange.

1. Knowledge Mining

2. Natural Language Processing

3. Computer vision

4. Anomaly Detection

5. Conversational AI

10. When training a model why should you randomly split the rows into separate subsets?

1. To train multiple models simultaneously to attain better performance

2. To train the model twice to achieve better accuracy

3. To test the model by using data that was not used to train the model

11. Which service has the capability to detect printed and handwritten text in images?

1. Text Analytics

2. Computer Vision- OCR

3. Translator Text

12. Keyphrase extraction can be used to retrieve all city names in a text string.

1. No

2. Yes

13. You have a lecture note which is sorted as text. For your exam, you need to extract key terms from the notes to generate summaries. Which type of AI workload should you use?

1. Anomaly detection

2. Conversational AI

3. Natural language processing

4. Computer vision

14. You are building a Language Understanding model for an e-commerce business. You need to ensure that the model detects when utterances are outside the intended scope of the model. What should you do?

1. Create a prebuilt task entity

2. Create a new model

3. Add utterances to the None intent

4. Test the model by using new utterances

15. A separate chatbot is required for each communication channel.

1. No

2. Yes

16. Select the most appropriate Natural Language Processing workload for below scenario. Scenario: Translate email communication to a specific language.

1. Language modelling

2. Translation

3. Sentiment analysis

4. Speech recognition and speech synthesis

5. Entity recognition

6. Key phrase extraction

17. Smart home devices that answer questions like “What is the weather today?“ are an example of natural language processing.

1. Yes

2. No

18. From the listed scenarios, select correct scenario where Natural language processing can be used

1. Stop a process in a factory when extremely high temperatures are registered.

2. Classify email messages as work-related or personal.

3. Predict the number of future car rentals.

4. Predict which website visitors will make a transaction.

19. Choose an answer and complete the sentence correctly. Identifying segments of your customer base using Recency, Frequency, Monetary (RFM) values ?? is [ ].

1. This is an example of regularization.

2. This is an example of classification.

3. An example of regression.

4. This is an example of clustering.

20. “Answering frequently asked questions“ is a scenario of Conversational AI.

1. No

2. Yes

21. In a group of people,  _______ facial recognition task can able to point out, who the person is.

1. Grouping

2. Verification

3. Similarity

4. Identification

22. You plan to build an application that uses the Speech service to transcribe audio recordings of phone calls into text. You also plan to submit the transcribed text to the Text Analytics service to extract key phrases. Moreover, you want to manage access and billing for the application services in a single Azure resource. Which type of Azure resource should you create?

1. Speech

2. Cognitive Services

3. Text Analytics

4. Custom Vision

23. Web chatbots can interact with users visiting a website.

1. Yes

2. No

24. “Providing an explanation of the outcome of a credit loan application“ is an example of the Microsoft transparency principle for responsible AI.

1. No

2. Yes

25. Which scenario is an example of webchat bot?

1. Translate into English questions entered by customers at a kiosk so that the appropriate person can call the customer

2. Determine whether reviews entered on a website for a concert are positive or negative

3. From a website interface answer common questions about the  scheduled events and ticket purchases for a music festival

4. Accepts questions through email and then route the email messages to the correct person based on the content of the message

26. Pair the facial recognition task with appropriate questions. Whether the two face images are of the same person.

1. Grouping

2. Verification

3. Similarity

4. Identification

27. Automated machine learning is the process of automating time-consuming and repetitive tasks in machine learning model development.

1. No

2. Yes

28. You need to develop a mobile app that allows your employees to scan and save expenses while traveling. What Kind of Computer Vision Should you Use?

1. Object detection

2. Optical Character Recognition (OCR)

3. Image classification

4. Face detection

29. If you create a Cognitive Services resource, client applications need different keys and endpoint for different service they use.

1. Yes

2. No

30. Language services can be used to extract key entities from phone transcripts.

1. No

2. Yes

31. You can detect the location of content in an image by creating an object detection model with the Custom Vision service.

1. Yes

2. No

32. _____ can be used to build no-code apps that use built-in natural language processing models.Select the answer that correctly completes the sentence.

1. Microsoft Copilot Studio

2. Microsoft Bot Framework

3. Azure Health Bot

33. Which machine learning algorithm is mostly used for predicting the values of categorical variables.

1. Logistic regression

2. K-Means

3. Linear regression

34. Using RFM ( Recency, Frequency, Monetary) values to identify segments of a customer base is an example of _____Select the answer that correctly completes the sentence.

1. Regularization

2. Clustering

3. Classification

4. Regression

35. Which of the provided options is considered the calculated probability of a correct image classification?

1. Accuracy

2. Sentiment

3. Root Mean Square Error

4. Confidence

36. You have published your Language Understanding application. What information does a client application developer need to get predictions from it?

1. The Azure credentials of the user who published the Language Understanding application

2. The endpoint and key for the application's prediction resource

3. The endpoint and key for the application's authoring resource

37. You need to deploy a real-time inference pipeline as a service for others to consume. Select the appropriate option to where you must deploy the model.

1. From Azure machine learning designer to Azure Kubernetes service

2. From Azure machine learning designer to Local web service

3. From Azure machine learning designer to Azure machine learning compute.

4. From Azure machine learning designer to Azure container instances

38. You have created a training pipeline in azure machine learning designer and now you want to experiment with the pipeline/code for which you need scalable clusters of virtual machines. What kind of compute target is suitable for the task mentioned.

1. Inference Clusters

2. Compute Clusters

3. Compute Instances

39. You need to convert handwritten notes into digital text. Which type of computer vision should you use?

1. Object detection

2. Image classification

3. Facial detection

4. Optical Character Recognition (OCR)

40. You are working as a solution architect for an online retail store and have been tasked to create a service that will go through all the reviews of a product and should be able to detect if a customer is happy with the product or upset. Which type of AI workload should you use?

1. Natural language processing

2. Semantic segmentation

3. Regression

4. Anomaly detection

41. Which of the following Workload type matches the below workloadAn automated chat to answer questions about refunds and exchanges on an E-commerce site.

1. Conversational AI

2. Anomaly detection

3. Computer Vision

4. Sentiment Analysis

42. Which of the following results does an object detection model typically return for an image?

1. Bounding box coordinates that indicate the region of the image where all of the objects it contains are located

2. A class label, probability, and bounding box for each object in the image

3. A class label and probability score for the image

43. Azure Bot Service allows you to import Frequently Asked Questions (FAQs) directly into your question and answer sets.

1. Yes

2. No

44. You are building an AI system. What tasks should be included to ensure that the service meets Microsoft‘s Principles of Responsible AI Transparency?

1. Have all visuals have associated text that can be read by screen readers.

2. Enable autoscaling so that your service scales on demand.

3. Provide documentation to help developers debug their code.

4. Ensure your training dataset is representative of the population.

45. Predicting how many vehicles will travel across a bridge on a given day is an example of what type of machine learning?

1. Regression

2. Classification

3. Clustering

46. You are building an application which will able to tell if the person looks like other people. Which facial recognition tasks will be appropriate for the above scenario?

1. Identification

2. Similarity

3. Verification

4. Grouping

47. You have Microsoft_Employee1 dataset having 16 rows and 4 columns. You have another dataset Microsoft_Employee2 having 16 rows and 3 columns. You want to combine these 2 datasets in such a way that the resultant dataset will have 16 rows and 7 columns. You dragged these 2 datasets in the azure machine learning designer canvas. Which module will help to achieve the desired result?

1. Drag the Select columns in Dataset module into the canvas and connect the output of Microsoft_Employee1 and Microsoft_Employee2 as input.

2. Drag the Remove duplicate Rows module into the canvas and connect the output of Microsoft_Employee1 and Microsoft_Employee2 as input.

3. Drag the Add Rows module into the canvas and connect the output of Microsoft_Employee1 and Microsoft_Employee2 as input.

4. Drag the Add Columns module into the canvas and connect the output of Microsoft_Employee1 and Microsoft_Employee2 as input.

48. You want to use the Computer Vision service to identify the location of individual items in an image. Which of the following features should you retrieve?

1. Categories

2. Objects

3. Tags

49. You need to provision an Azure resource that will be used to author a new Language Understanding application. What kind of resource should you create?

1. Text Analytics

2. Language Understanding

3. Cognitive Services

50. You have been assigned a task to extract key terms from insurance claim reports, that are stored as text, and generate summaries. Which type of AI workload should you use?

1. Anomaly detection

2. Natural language processing

3. Computer vision

4. Conversational AI

FAQs

1. What is the Microsoft Azure AI Fundamentals (AI-900) certification?

The AI-900 certification validates your foundational knowledge of artificial intelligence (AI) and machine learning (ML) concepts, and how they are implemented using Microsoft Azure services.

2. How do I become certified in Azure AI Fundamentals (AI-900)?

You must pass the AI-900 exam, which tests your understanding of AI workloads, ML principles, and Azure AI tools like Cognitive Services and Azure Machine Learning.

3. What are the prerequisites for the Microsoft Azure AI-900 exam?

There are no prerequisites for the AI-900 exam. It is designed for beginners and professionals starting their journey in AI or cloud computing.

4. How much does the Azure AI Fundamentals certification cost?

The AI-900 exam costs $99 USD, though prices may vary by country or region.

5. How many questions are in the Microsoft AI-900 exam?

The exam typically includes 40–60 multiple-choice and scenario-based questions, with a 60-minute duration.

6. What topics are covered in the Azure AI Fundamentals (AI-900) certification exam?

It covers AI workloads, computer vision, natural language processing (NLP), conversational AI, and ML fundamentals on Azure.

7. How difficult is the Microsoft Azure AI Fundamentals (AI-900) exam?

It’s considered an entry-level exam, suitable for candidates with no prior coding or AI experience.

8. How long does it take to prepare for the AI-900 certification exam?

Most candidates prepare in 2–4 weeks, depending on their background in cloud and AI concepts.

9. What jobs can I get after earning the Azure AI Fundamentals certification?

You can work as an AI Analyst, Data Technician, Cloud Support Associate, or Junior AI Engineer.

10. How much salary can I earn with the Microsoft Azure AI Fundamentals (AI-900) certification?

Professionals with AI-900 certification typically earn between $70,000–$100,000 per year, depending on role and experience.