Study guide
1.3 Identifying Common AI Workloads and Use Cases
Microsoft Azure AI Fundamentals AI-900
Track your progress as you read
Finished sections get a check mark.
1 Introduction
About the Azure AI Fundamentals (AI-900) Exam Study Strategies and Exam Tips for the AI-900
2 Domain 1: Describe AI Workloads and Considerations (15-20%)
1.1 What Is Artificial Intelligence? 1.2 Responsible AI Principles 1.3 Identifying Common AI Workloads and Use Cases 1.4 Responsible AI Tools and Practices on Azure
3 Domain 2: Fundamental Principles of Machine Learning on Azure (20-25%)
2.1 Core Machine Learning Concepts 2.2 Regression Models 2.3 Classification Models 2.4 Clustering Models 2.5 Deep Learning and Neural Networks 2.6 Azure Machine Learning Service
4 Domain 3: Computer Vision Workloads on Azure (15-20%)
3.1 Computer Vision Concepts and Tasks 3.2 Azure AI Vision Service 3.3 Azure AI Custom Vision and Face Services 3.4 Azure AI Document Intelligence
5 Domain 4: Natural Language Processing Workloads on Azure (15-20%)
4.1 NLP Concepts and Common Tasks 4.2 Azure AI Language Service 4.3 Azure AI Speech Service 4.4 Azure AI Translator Service
6 Domain 5: Generative AI Workloads on Azure (15-20%)
5.1 Generative AI Fundamentals 5.2 Azure OpenAI Service 5.3 Prompt Engineering 5.4 Retrieval-Augmented Generation (RAG) 5.5 Responsible Generative AI 5.6 Microsoft Copilot and AI-Powered Applications
7 Exam Review and Full-Length Practice Questions
6.1 Domain Summary and Key Comparisons 6.2 Practice Questions: AI Workloads and Responsible AI 6.3 Practice Questions: Machine Learning 6.4 Practice Questions: Computer Vision, NLP, and Generative AI
1.3 Identifying Common AI Workloads and Use Cases
Key Takeaways Explain This Topic AI
Prediction workloads use regression to forecast continuous values (prices, temperatures, quantities) from historical data.
Anomaly detection identifies unusual patterns in data — fraud detection, equipment failure prediction, and quality control are common use cases.
Computer vision workloads extract information from visual data — classification, detection, OCR, and facial analysis are the key subtypes.
Conversational AI combines NLP and generative AI to create chatbots and virtual assistants that interact naturally with users.
The exam tests your ability to match real-world scenarios to the correct AI workload type — focus on identifying the input data type and desired output.
Last updated: October 2026
Quick Answer: The AI-900 exam expects you to identify seven common AI workload types: prediction/forecasting, classification, anomaly detection, computer vision, NLP, conversational AI, and generative AI. The key to identifying the correct workload is analyzing the input data and desired output in the scenario.
Workload Identification Framework
When the exam presents a business scenario, use this decision tree:
Step 1: What is the input?
- Images or video → Computer Vision
- Text → NLP or Generative AI
- Speech/audio → NLP (Speech) or Generative AI
- Numerical/tabular data → Machine Learning (regression, classification, or clustering)
- Time-series data → Anomaly Detection or Prediction
Step 2: What is the desired output?
- A continuous number (price, temperature, quantity) → Regression
- A category or label → Classification
- Groups of similar items → Clustering
- Unusual patterns flagged → Anomaly Detection
- New content (text, images, code) → Generative AI
- Understanding of text meaning → NLP
- Understanding of visual content → Computer Vision
- Natural conversation → Conversational AI
- Detailed Workload Types with Use Cases
- Prediction and Forecasting (Regression)
Regression models predict continuous numerical values based on historical data and input features.
- Use Case Input Features Predicted Output
- House price prediction Size, location, bedrooms, age Price in dollars
- Sales forecasting Historical sales, season, promotions Future sales volume
- Temperature prediction Date, location, historical weather Temperature in degrees
- Energy consumption Time of day, season, building size Kilowatt-hours
- Delivery time estimation Distance, traffic, weather Minutes until delivery
- Classification
Classification models assign items to predefined categories based on their features.
- Use Case Input Categories
- Email spam detection Email content, sender, metadata Spam / Not Spam
- Medical image diagnosis X-ray or MRI image Disease / No Disease
- Credit risk assessment Income, credit history, employment Approve / Deny
- Product defect detection Product image Defective / Not Defective
- Customer churn prediction Usage patterns, payment history Churn / Stay
- Anomaly Detection
Anomaly detection identifies data points that deviate significantly from expected patterns.
- Use Case Normal Pattern Anomaly
- Credit card fraud Typical spending patterns Unusual large purchase in a foreign country
- Equipment monitoring Normal vibration and temperature Sudden spike in vibration before failure
- Network security Normal traffic patterns Spike in traffic from unknown source (DDoS)
- Quality control Consistent product dimensions Product outside acceptable tolerance
- Healthcare monitoring Normal heart rhythm Irregular heartbeat detected by wearable
- Computer Vision
Computer vision extracts information from images and videos.
- Task Description Example
- Image classification Assign a label to an entire image "This is a photo of a cat"
- Object detection Identify and locate objects with bounding boxes "There is a car at position (x, y, w, h)"
- Semantic segmentation Classify every pixel in an image Autonomous driving — road, sidewalk, car, pedestrian
- OCR Extract text from images Reading text from a scanned receipt
- Facial detection Detect faces and their attributes Locating faces, landmarks, and head pose
- Facial recognition Identify or verify a specific person Unlocking a phone with Face ID
- Natural Language Processing (NLP)
NLP understands, interprets, and generates human language.
- Task Description Example
- Sentiment analysis Determine if text is positive, negative, or neutral "Great product!" → Positive
Key phrase extraction Identify main topics in text "The hotel room was clean and spacious" → "hotel room", "clean", "spacious"
Named entity recognition Identify entities (people, places, dates) "John visited Paris on March 5" → Person: John, Location: Paris, Date: March 5
- Language detection Identify the language of text "Bonjour le monde" → French
- Text summarization Condense long text into key points Summarize a 10-page article into 3 sentences
- Speech-to-text Convert spoken words to written text Voice dictation
- Text-to-speech Convert written text to spoken audio Screen reader for visually impaired users
- Translation Convert text from one language to another English → Spanish
- Conversational AI
Conversational AI builds intelligent chatbots and virtual assistants that interact naturally with users.
- Use Case Description
- Customer service chatbot Answer FAQs, process orders, handle complaints
- IT help desk bot Reset passwords, troubleshoot issues, create tickets
- Healthcare triage bot Assess symptoms, recommend next steps, schedule appointments
- E-commerce assistant Product recommendations, order tracking, returns
- Internal knowledge bot Answer employee questions from company documents
- Generative AI
Generative AI creates new content based on prompts or input data.
- Use Case Input Output
- Content writing "Write a product description for..." Marketing copy
- Code generation "Write a Python function that..." Working code
- Image generation "A sunset over a mountain lake" Generated image
- Summarization Long document Concise summary
- Translation Text in English Text in Japanese
- Q&A over documents Question + context documents Grounded answer
On the Exam: The most common question pattern is: "A company needs to [business scenario]. Which type of AI workload is this?" Identify the input type and desired output to determine the correct workload. If the scenario involves creating new content, it is generative AI. If it involves understanding existing content, it is NLP or computer vision.