AI-901
Free Microsoft Certified: Azure AI Fundamentals Practice Test
12 exam-style questions with full explanations — no sign-up. Score yourself, then close your gaps with the full course.
Exam fee ~$992 exam domainsLevel Beginner2 timed practice tests in the course
✅ Free practice — no sign-up📝 Real exam-style questions💡 Detailed explanations💸 30-day money-back via Udemy
A healthcare provider wants to build a solution that analyzes patient X-ray images to detect potential abnormalities and flag cases for radiologist review. Which AI workload does this scenario BEST represent?
Computer vision workloads analyze and interpret visual content including images. Detecting abnormalities in X-rays requires image classification and object detection capabilities, which are core computer vision tasks. Azure AI Vision and Custom Vision services support medical imaging scenarios.
A developer is building a customer service chatbot using Azure AI Foundry. The chatbot must maintain conversation context across multiple turns and provide consistent responses based on company policies stored in a knowledge base. Which Azure AI Foundry component should the developer implement?
A single-agent solution with RAG is specifically designed for this scenario. It maintains conversation context, grounds responses in the company knowledge base, and provides consistent answers by retrieving relevant information before generating responses. This is the core pattern for building knowledge-based chatbots in Azure AI Foundry.
A company deploys an AI hiring assistant that screens resumes. During testing, the system consistently scores candidates from certain universities lower than others, even when qualifications are similar. Which Responsible AI principle is MOST directly violated?
Fairness requires AI systems to treat all people equitably and not discriminate based on protected characteristics or proxies for them. University affiliation can be a proxy for socioeconomic status, race, or geography. Systematically scoring candidates differently based on their university violates fairness principles regardless of actual qualifications.
A company needs to analyze customer feedback emails to determine overall sentiment and extract key topics mentioned. The solution must identify whether feedback is positive, negative, or neutral, and extract specific product names and issues. Which Azure AI service combination provides this functionality?
Azure AI Language provides both sentiment analysis (positive/negative/neutral classification) and key phrase extraction (identifying product names and issues) in a single service. This directly addresses both requirements stated in the scenario.
An organization wants to build a chatbot that answers customer questions by retrieving information from their product documentation and support articles. The chatbot should provide accurate answers grounded in company knowledge. Which approach BEST addresses the risk of hallucination?
Retrieval-augmented generation (RAG) combines information retrieval with generation. It first retrieves relevant documents from the company's knowledge base, then uses that retrieved context to ground the LLM's response. This significantly reduces hallucination because answers are based on actual source material rather than the model's training data. Azure AI Search integrated with Azure OpenAI Service supports RAG patterns.
A developer deploys a GPT-4 model in Azure AI Foundry portal and writes the following system prompt: 'You are a helpful assistant.' User messages frequently receive generic responses that don't align with the company's technical support style. What is the BEST way to improve response quality?
Writing effective system prompts is a core AI-901 skill. The system prompt should include specific role definition, tone/style guidelines, and examples of desired response format. This is the standard approach to align model behavior with organizational requirements without changing the model or parameters unnecessarily.
A retail company wants to automatically extract product names, quantities, and prices from supplier invoices in various formats (PDF, scanned images, emails). Which AI workload is MOST appropriate for this scenario?
Document intelligence (formerly Form Recognizer) is specifically designed to extract structured information from documents including invoices. Azure AI Document Intelligence includes prebuilt models for invoices that extract key-value pairs, line items, tables, and specific fields like vendor, total, and dates. It handles multiple formats and layouts automatically.
An application needs to convert recorded customer support calls to text for analysis and quality assurance. The audio files are in WAV format, average 10 minutes in length, and contain multiple speakers. Which Azure AI service should be used?
Azure AI Speech speech-to-text is the dedicated service for converting audio to text. It handles various audio formats including WAV, supports speaker diarization for multiple speakers, and is optimized for speech recognition tasks.
A developer is building a text summarization feature using Azure OpenAI Service. The input document is 15,000 words. The model has a context window of 8,192 tokens. Assuming an average of 1.3 tokens per word, what is the BEST approach to handle this document?
15,000 words × 1.3 tokens/word = 19,500 tokens, which exceeds the 8,192 token context window. Chunking the document into smaller segments (each under ~6,000 tokens to leave room for the prompt and output), summarizing each chunk, then combining or further summarizing the results is the standard approach. This is a common pattern in document processing with LLMs.
A retail company wants to automatically categorize product images uploaded by sellers into predefined categories such as 'Electronics', 'Clothing', 'Home & Garden', and 'Sports'. The solution must work without training a custom model. Which Azure AI capability should be implemented?
Azure Computer Vision provides pre-trained image classification models that can categorize images into standard categories without requiring custom training. This matches the requirement to work 'without training a custom model'.
A financial services company is deploying an AI agent that provides investment advice to customers. The agent can execute trades, retrieve account balances, and access market data through API calls. Which Responsible AI principle requires that customers can easily understand what data the agent accesses and what actions it can perform on their behalf?
Transparency requires that AI systems be understandable and that their capabilities, limitations, and decision-making processes be clear to users. For an AI agent with significant permissions (executing trades, accessing financial data), users must understand what the agent can and cannot do, what data it accesses, and how it operates. This enables informed consent and appropriate trust calibration.
A legal firm needs to extract specific fields (contract date, party names, clauses) from thousands of scanned contract PDFs with varying layouts. The documents are unstructured and not in standard forms. Which Azure AI service is BEST suited for this task?
Azure Content Understanding is specifically designed for extracting structured information from unstructured documents, forms, and content with varying layouts. It handles the complexity of different document formats and intelligently extracts fields without requiring standard templates.
Ready for the real thing?
The full course has two full-length practice tests, video lessons for every exam domain, hands-on labs and detailed answer explanations.
Start my full course on Udemy →AI-901 exam — quick answers
How much does the AI-901 exam cost?
The exam fee is approximately $99 and varies by region — confirm current pricing with the certification vendor before you book.
What topics are on the exam?
It covers 2 domains: Identify AI concepts and responsibilities (40-45%), Implement AI solutions by using Microsoft Foundry (55-60%). The full course has a dedicated chapter, lab and practice-test coverage for each.
Is this practice test really free?
Yes — all questions on this page are free with explanations and no sign-up. The paid Udemy course adds two full-length timed exams, video lessons and hands-on labs.
Will this prepare me for the real exam?
The questions mirror the real exam's style and are mapped to the official domains. This is exam-focused preparation — combine the free test with the full course's timed simulations to gauge your readiness.