Microsoft AI-900: Microsoft Azure AI Fundamentals

Last checked October 22, 2026 against the official Microsoft exam page.

1. Question 1Describe features of generative AI workloads on Azure

   What is a large language model mainly used for in generative AI workloads?

   1. ARouting network packets between the segments of a corporate network according to the rules configured by the team
   2. BGenerating and summarizing natural language text, such as drafts, answers, and summaries, from a prompt it receives
   3. CPrinting invoices on paper stock in the accounts department each month for the customers of the business units
   4. DDetecting faces in photographs so that the images can be sorted by person in the photo library of the company
   Show answer and explanationHide answer

   Correct answer: B

   Large language models generate and transform text in response to prompts, which covers drafting, answering, and summarizing. Packet routing, face detection, and invoice printing are not what these models do.

2. Question 2Describe features of Natural Language Processing (NLP) workloads on Azure

   Which NLP feature extracts the main talking points from a passage of text?

   1. ASpeech synthesis, which converts written text into a spoken audio output that a listener can hear
   2. BKey phrase extraction, which identifies the main concepts and talking points that a passage discusses
   3. CImage classification, which assigns category labels to pictures so that they can be filtered and searched
   4. DDemand forecasting, which predicts the number of units that will be sold over a future period of time
   Show answer and explanationHide answer

   Correct answer: B

   Key phrase extraction returns the main concepts of a text, which is how talking points are identified. Speech synthesis, image classification, and forecasting work on different types of input or produce different outputs.

3. Question 3Describe fundamental principles of machine learning on Azure

   Which machine learning technique predicts a continuous numeric value, such as the price of a house?

   1. AClustering, which groups unlabelled items into sets based on how similar their features are to one another
   2. BClassification, which assigns each item to one of a fixed set of categories that are defined in advance
   3. CRegression, which predicts a continuous numeric value from the input features of each example in the data
   4. DReinforcement learning, which learns a policy through rewards and penalties received after taking actions
   Show answer and explanationHide answer

   Correct answer: C

   Regression predicts continuous values, which is what a house price is. Clustering groups without labels, classification predicts categories, and reinforcement learning learns from rewards rather than predicting a value directly.

4. Question 4Describe features of computer vision workloads on Azure

   Which computer vision feature assigns one or more labels to an entire image?

   1. AObject detection, which locates each object in an image and returns a bounding box around each one found
   2. BFace detection, which locates human faces in an image and returns their positions within the picture frame
   3. COptical character recognition, which extracts printed or handwritten text from an image and returns it as text
   4. DImage classification, which assigns one or more category labels to the image as a whole, such as cat or dog
   Show answer and explanationHide answer

   Correct answer: D

   Image classification labels the whole image, while object detection locates objects, OCR extracts text, and face detection finds faces. Only image classification matches the question as asked.

5. Question 5Describe Artificial Intelligence workloads and considerations

   Which AI workload is best suited to extracting printed text from scanned images of documents?

   1. AOptical character recognition, a computer vision capability that reads printed text from images and scans
   2. BNatural language processing, which analyses the meaning of text that has already been typed into a system
   3. CAnomaly detection, which flags unusual values in a series of numbers recorded over time for a process
   4. DForecasting, which predicts future numeric values from historical data, such as next quarter's demand
   Show answer and explanationHide answer

   Correct answer: A

   Optical character recognition is a computer vision workload that converts printed text in images into machine-readable text. Natural language processing works on text that already exists, while anomaly detection and forecasting work on numeric series.