AI-901: Azure AI Fundamentals Practice Exam

What the AI-901 exam covers

• Identify AI concepts and capabilities326 questions
• Implement AI solutions by using Microsoft Foundry387 questions

Free AI-901 sample questions

A sample of 10 questions with answers and explanations. Sign up free to practice all 713.

1. Question 1Identify AI concepts and capabilities

   In supervised machine learning, what is the role of a label in a training dataset?

   • AThe known output value the model is trained to predictCorrect
   • BThe input variable the model uses to make a prediction
   • CThe numeric weight assigned to each neuron in a network
   • DThe score that measures how accurate a prediction was
   ✓ Correct answer: A

   In supervised learning each training example pairs input features with a label, which is the correct answer for that example. The model adjusts itself to map features to labels, and after training it predicts labels for new, unlabeled data.

   Why the other options are wrong

   • BThat describes a feature, which is an input variable, not the target output.
   • CWeights are internal model parameters learned during training, not labels in the dataset.
   • DThat describes an evaluation metric such as accuracy, not a label.
2. Question 2Identify AI concepts and capabilities

   A health app uses AI to analyze patient records and must ensure that the records are not exposed to unauthorized parties and cannot be extracted through attacks on the model. Which responsible AI principle is the primary focus?

   • APrivacy and securityCorrect
   • BTransparency
   • CInclusiveness
   • DFairness
   ✓ Correct answer: A

   Privacy and security require that AI systems protect personal data and remain secure against attacks and misuse. Preventing unauthorized exposure of patient records and guarding against extraction attacks are core to this principle. It covers both the confidentiality of data and the resilience of the system.

   Why the other options are wrong

   • BTransparency concerns explaining how the system works, not securing data.
   • CInclusiveness concerns serving diverse users, not protecting records.
   • DFairness concerns equitable treatment, not data protection.
3. Question 3Identify AI concepts and capabilities

   What is a copilot in the Microsoft generative AI context?

   • AAn AI assistant that helps users complete tasks using natural-language interaction, often grounded in specific dataCorrect
   • BA monitoring dashboard that visualizes response latency, token usage, and error rates for deployed AI endpoints
   • CA managed inference service that hosts, versions, and autoscales deployed models behind a REST endpoint
   • DA scheduled automation job that backs up virtual machine disks and databases to redundant cloud storage
   ✓ Correct answer: A

   A copilot is an AI-powered assistant that helps users perform tasks through natural-language conversation, frequently grounded in relevant business or application data. Copilots are commonly built on generative language models with grounding and prompt engineering. They aim to boost productivity by assisting rather than fully replacing the user.

   Why the other options are wrong

   • BA latency and usage dashboard monitors an endpoint's health; it does not converse with users or complete tasks.
   • CA managed inference endpoint hosts a model for scoring, but it is infrastructure, not a natural-language assistant.
   • DA backup job protects data on a schedule and has nothing to do with generative AI assistance.
4. Question 4Implement AI solutions by using Microsoft Foundry

   A team wants to reuse the same Azure Storage account and Azure AI Search resource across several related AI projects without reconfiguring each one. Where should these shared resources be configured?

   • AAs connections on the hub so all projects under it inherit themCorrect
   • BAs a separate resource group inside each project
   • CAs local files uploaded individually to every project
   • DAs environment variables set on the user's workstation
   ✓ Correct answer: A

   Connections defined at the hub level are available to all projects created under that hub, avoiding repeated setup. This is the intended pattern for sharing resources like Storage or AI Search across related projects. Projects can also define their own project-scoped connections when needed.

   Why the other options are wrong

   • BResource groups organize Azure resources, but they do not automatically share connections across Foundry projects.
   • CUploading files to each project does not create reusable connections to external services.
   • DWorkstation environment variables are local to one machine and do not share resources across cloud projects.
5. Question 5Implement AI solutions by using Microsoft Foundry

   A serverless deployment works in testing but a production client cannot connect. The endpoint URL and deployment name are confirmed correct. Which is the most likely remaining cause?

   • AThe production client is not sending a valid authentication key or tokenCorrect
   • BThe model needs to be retrained before production use
   • CServerless deployments cannot be used in production
   • DThe endpoint only accepts requests from the Foundry playground
   ✓ Correct answer: A

   If the URL and deployment name are correct but production still fails, the most common cause is that the production client lacks the valid key or token that the test environment had. Endpoints require authentication on every call. Verifying the credential and its permissions typically resolves the issue.

   Why the other options are wrong

   • BA working test deployment does not require retraining to serve production requests.
   • CServerless deployments are fully supported for production workloads.
   • DEndpoints accept requests from any authorized client, not only the playground.
6. Question 6Implement AI solutions by using Microsoft Foundry

   A deployed model in Microsoft Foundry starts returning HTTP 429 responses under heavy load. What is the most likely cause?

   • AThe deployment exceeded its allocated rate limit or quotaCorrect
   • BThe grounding index was deleted
   • CThe embedding model was set to the wrong language
   • DThe evaluation flow produced a low coherence score
   ✓ Correct answer: A

   An HTTP 429 Too Many Requests response indicates that traffic exceeded the deployment's allocated rate limit, such as its tokens-per-minute or requests-per-minute quota. To resolve it, teams can increase the quota, add capacity, or implement retry with backoff. This is a common capacity and cost management signal.

   Why the other options are wrong

   • BA deleted index would cause retrieval failures, not 429 rate-limit errors.
   • CA wrong embedding language would affect relevance, not produce 429 errors.
   • DA low coherence score is an evaluation result, not a runtime status code.
7. Question 7Implement AI solutions by using Microsoft Foundry

   A developer needs to inspect the intermediate steps an agent took during a run, such as which tool it called. What should they examine?

   • AThe agent's instructions field
   • BThe run steps associated with the runCorrect
   • CThe thread's creation timestamp
   • DThe model deployment's region setting
   ✓ Correct answer: B

   A run is composed of run steps that record the individual actions the agent performed, such as tool calls and message creation. Inspecting these steps lets a developer trace and debug the agent's behavior.

   Why the other options are wrong

   • AThe instructions describe intended behavior, not what happened in a run.
   • CA creation timestamp does not reveal the steps the agent took.
   • DThe deployment region has nothing to do with the run's actions.
8. Question 8Identify AI concepts and capabilities

   Which Image Analysis feature returns multiple descriptive sentences, each describing a different region of the same image?

   • ADense CaptionsCorrect
   • BCaption
   • CTags
   • DObjects
   ✓ Correct answer: A

   The Dense Captions feature generates a separate descriptive sentence for each of several regions within an image, giving more detailed coverage than a single caption. Each region caption is associated with a bounding box.

   Why the other options are wrong

   • BCaption returns just one sentence for the whole image, not per-region descriptions.
   • CTags returns individual words, not sentences describing regions.
   • DObjects returns labels and boxes without descriptive sentences.
9. Question 9Identify AI concepts and capabilitiesSelect all that apply

   Which of the following are valid uses of Azure AI speech-to-text? (Choose two.)

   • ATranscribing a recorded podcast episode into textCorrect
   • BGenerating live captions from a streaming lectureCorrect
   • CSynthesizing an announcement in a neural voice
   • DRotating and cropping an uploaded photo
   ✓ Correct answer: A, B

   Speech-to-text transcribes recorded audio such as a podcast and can also produce live captions from streaming audio. Both convert speech into text. Synthesizing announcements is text-to-speech and photo editing is an image task.

   Why the other options are wrong

   • CSynthesizing an announcement is text-to-speech, which produces audio rather than transcribing it.
   • DRotating and cropping a photo is an image-processing task unrelated to speech.
10. Question 10Implement AI solutions by using Microsoft FoundrySelect all that apply

    Which of the following are text-analysis capabilities provided by Azure AI Language that you can call from a lightweight app? (Choose three.)

    • ASentiment analysisCorrect
    • BKey phrase extractionCorrect
    • CLanguage detectionCorrect
    • DRemoving the background from a product photo
    • EDetecting the faces in a group photo
    ✓ Correct answer: A, B, C

    Azure AI Language offers sentiment analysis, key phrase extraction, language detection, named entity recognition, PII detection, and summarization for text. Background removal and face detection are image tasks handled by Azure AI Vision and Face, not Language.

    Why the other options are wrong

    • DBackground removal is an Azure AI Vision image feature.
    • EFace detection is provided by Azure AI Vision / Face, not Language.

Who this AI-901 practice exam is for

This practice set is for anyone preparing for the AI-901: Azure AI Fundamentals exam at the foundational level - from first-time candidates building a foundation to experienced Microsoft practitioners doing a final review before test day. If you learn best by working through realistic questions and reading why each answer is right or wrong, it is built for you.

How to use this AI-901 practice exam

1. Start with the free sample questions above to gauge your current baseline.
2. Read the full explanation on every question, including why each wrong option is wrong.
3. Track your weak domains and focus your study where you are losing the most marks.
4. Once you are scoring consistently well, take a timed, full-length mock exam.
5. Use your readiness score to decide when you are ready to book the real AI-901 exam.

Related Microsoft resources

• AI-901 study guideKey concepts
• Microsoft practice examsAll Microsoft
• Certification pathWhere this fits
• Certification exam guides & tipsBlog
• Plans & pricingFree & paid
• SC-200 practice examRelated
• AZ-140 practice examRelated
• AZ-900 practice examRelated

AI-901 practice exam FAQ