What is the AI-900 Certification?
Microsoft Azure AI Fundamentals (AI-900) is the entry-level AI certification from Microsoft. It validates your understanding of AI and machine learning concepts and how they are implemented in Azure AI services.
You do not need to write code for this exam. It's designed for both technical and non-technical professionals who want to understand AI in a cloud context.
Exam facts: - Cost: $165 USD - Passing score: 700 out of 1000 - Format: 40 to 60 questions - Duration: 45 minutes - Best paired with: AZ-900
Exam Domains
| Domain | Weight |
|---|---|
| AI and Machine Learning Concepts | 20 to 25% |
| Computer Vision Workloads on Azure | 15 to 20% |
| Natural Language Processing on Azure | 15 to 20% |
| Document Intelligence and Knowledge Mining | 10 to 15% |
| Generative AI Workloads on Azure | 15 to 20% |
| Responsible AI Principles | 10 to 15% |
Domain 1: AI and Machine Learning Concepts
What is Artificial Intelligence? AI is the simulation of human intelligence processes by computers, including learning, reasoning, and problem-solving.
Machine Learning vs Deep Learning: - Machine Learning: algorithms that learn patterns from data without being explicitly programmed - Deep Learning: a subset of ML using neural networks with many layers — used for images, speech, text
Types of Machine Learning: - Supervised Learning: Trained on labelled data. Types: - Classification: predicts a category (spam/not spam, disease/healthy) - Regression: predicts a number (house price, temperature) - Unsupervised Learning: No labels. Types: - Clustering: groups similar items (customer segmentation) - Anomaly detection: finds unusual patterns (fraud detection) - Reinforcement Learning: Agent learns by trial and error, receiving rewards
Key concepts: - Feature: an input variable used to make predictions - Label: the output value being predicted - Model: the algorithm trained on data - Training data vs Test data - Overfitting vs Underfitting
Domain 2: Computer Vision on Azure
Computer Vision is the ability of AI to interpret and understand visual information.
Common computer vision tasks: - Image Classification: What is in this image? (cat, dog, car) - Object Detection: Where are the objects in this image? (bounding boxes) - Semantic Segmentation: Identify the exact pixels of each object - Facial Recognition: Identify or verify individuals - Optical Character Recognition (OCR): Extract text from images
Azure Computer Vision services: - Azure AI Vision: Image analysis, object detection, OCR (Read API) - Azure Face: Face detection, verification, and identification - Azure Video Analyzer: Analyse and index video content - Azure Custom Vision: Train your own image classification model
Domain 3: Natural Language Processing on Azure
NLP is the ability of AI to understand and generate human language.
Common NLP tasks: - Sentiment analysis: Is this review positive, negative, or neutral? - Key phrase extraction: What are the main topics in this text? - Language detection: What language is this written in? - Named entity recognition: Who, what, and where are mentioned? - Text translation: Translate between languages - Text summarisation: Create a concise summary
Azure NLP services: - Azure AI Language (formerly Text Analytics + LUIS + QnA Maker) - Azure AI Translator: Real-time text translation - Azure AI Speech: Speech to text, text to speech, speaker recognition - Azure Cognitive Search: AI-powered search with knowledge mining
Domain 4: Generative AI on Azure
This is the newest and fastest-growing exam domain.
What is Generative AI? AI systems that can create new content — text, images, code, audio — based on prompts.
Key concepts: - Large Language Models (LLMs): Trained on massive text datasets (GPT-4, Claude, Gemini) - Prompt Engineering: Crafting inputs to get the best outputs from AI - Tokens: How LLMs process text (roughly 1 token per word) - Temperature: Controls creativity vs predictability of AI outputs - Retrieval-Augmented Generation (RAG): Giving LLMs access to custom knowledge bases - Hallucination: When AI confidently states incorrect information
Azure Generative AI services: - Azure OpenAI Service: Access to GPT-4, DALL-E, Whisper via Azure - Microsoft Copilot: AI assistant built into Microsoft 365 products - Azure AI Foundry (formerly Azure AI Studio): Platform for building AI apps
Domain 5: Responsible AI
Microsoft's six Responsible AI principles appear on every AI-900 exam. Memorise them:
| Principle | What it means |
|---|---|
| **Fairness** | AI should treat all people equitably and not discriminate |
| **Reliability and Safety** | AI should perform reliably and safely |
| **Privacy and Security** | AI should be secure and respect privacy |
| **Inclusiveness** | AI should empower everyone, including people with disabilities |
| **Transparency** | AI systems should be understandable and explainable |
| **Accountability** | People should be accountable for AI systems |
Memory tip: FRIPPTA (Fairness, Reliability, Inclusiveness, Privacy, Privacy, Transparency, Accountability)
Best Free Resources for AI-900
- Microsoft Learn AI-900 learning path (free, official)
- Azure AI Foundry and Vision Studio: Free hands-on demos
- Adam Marczak's YouTube AI-900 playlist: Clear and well-paced
- Microsoft Official Practice Assessment: Free on Microsoft's site
Key Exam Strategy
Many AI-900 questions are scenario-based: "A company wants to do X — which Azure AI service should they use?" Build a mental map of which service does what: - Extract text from images: Azure AI Vision (Read API) - Analyse sentiment in customer reviews: Azure AI Language - Build a FAQ chatbot: Azure AI Language (Question Answering) - Translate documents: Azure AI Translator - Generate images from text: Azure OpenAI (DALL-E)