Practice Exam AI-901: Azure AI Fundamentals

Question bank: 100 · timed mock exam: 45 min

Official exam page on the vendor's site (opens in a new tab) Official exam guide (opens in a new tab) Official prep course (opens in a new tab)

Exam guide: format, domains and how to prepare

Compare: AI-901 vs AIF-C01

Compare: AI-901 vs AI-103

Compare: AZ-900 vs AI-901

  • Implement AI solutions by using Microsoft Foundry 58 questions
  • Identify AI concepts and capabilities 42 questions
Study mode (free) Exam mode (subscribers)

Free: 25 questions in study mode, all domains, no account.

Subscription: the 100 questions of this certification, the timed exam mode, and all other certifications too.

Subscribe · €9.99 / month

Secure card payment (Stripe) · No commitment · Cancel online

How to prepare for Azure AI Fundamentals with InfraPrep

The 100 questions cover 2 domains: Implement AI solutions by using Microsoft Foundry, Identify AI concepts and capabilities. 25 of them are free, without an account.

Study mode

Answer at your pace: each answer is corrected straight away, with an explanation and the link to the documentation it relies on.

Exam mode

A timed exam of 45 minutes, with a question map to flag and come back to questions before you submit.

Results by domain

Your score is broken down by domain, so you know what to review first.

Original questions

Every question is written from the public documentation, never copied from the exam. You can contest an answer from study mode.

10 free sample questions

Question 1 Identify AI concepts and capabilities

A recruitment firm tests its résumé screening model and finds that candidates with the same qualifications receive lower scores when they belong to an older age group. Which Microsoft responsible AI principle is most directly at risk?

  1. Fairness
  2. Transparency
  3. Accountability
  4. Privacy and security
Show the answer and the explanation

Correct answer: A

The fairness principle says AI systems should treat everyone fairly and give the same recommendations to people with similar qualifications, which is exactly what fails here.

Source: official documentation

Question 2 Implement AI solutions by using Microsoft Foundry

A developer has just deployed a chat model in Microsoft Foundry. Before writing any code, she wants to try system prompts and see how changing temperature, top_p, and max_tokens affects the answers. Which Foundry feature is intended for this?

  1. The model playground
  2. Content Understanding Studio
  3. The Azure Language quickstart
  4. The Speech CLI
Show the answer and the explanation

Correct answer: A

Foundry playgrounds provide an instant chat environment for prototyping where you can configure parameters such as temperature, top_p, and max_tokens and inject system prompts.

Source: official documentation

Question 3 Identify AI concepts and capabilities

An insurer uses an AI system to recommend whether claims should be denied. Company policy states that the system can never issue a final denial on its own: a named team signs off on every decision and answers for the outcome. Which responsible AI principle does this policy put into practice?

  1. Transparency
  2. Inclusiveness
  3. Accountability
  4. Reliability and safety
Show the answer and the explanation

Correct answer: C

Accountability means the people who design and deploy AI remain answerable for it, and that AI systems are not the final authority on decisions that affect people's lives.

Source: official documentation

Question 4 Implement AI solutions by using Microsoft Foundry

A team adds the line "Never disclose internal pricing" to the system message of its Foundry chat app and plans to treat this line as its only control against pricing leaks. What does the system message design guidance say about this plan?

  1. It is sufficient, if the line is repeated in every user message as well.
  2. It is insufficient, because a system message doesn't guarantee compliance.
  3. It is insufficient, because system messages are ignored by chat models.
  4. It is sufficient, because the system message is always obeyed by the model.
Show the answer and the explanation

Correct answer: B

The guidance states that a system message influences the model but doesn't guarantee compliance, so it should be layered with other mitigations such as filtering and evaluation.

Source: official documentation

Question 5 Identify AI concepts and capabilities

A regulator requires a bank to tell each applicant why its AI-assisted credit model approved or rejected their loan application. Which responsible AI principle does the bank address by providing these explanations?

  1. Fairness
  2. Reliability and safety
  3. Privacy and security
  4. Transparency
Show the answer and the explanation

Correct answer: D

Transparency covers helping people understand how AI decisions that affect their lives are made, and interpretability (useful explanations of system behavior) is a crucial part of it.

Source: official documentation

Question 6 Implement AI solutions by using Microsoft Foundry

A developer is writing a lightweight Python chat client with the Foundry SDK. He has created a project client from his project endpoint. According to the Foundry SDK guidance, how should the app send prompts to a deployed model?

  1. Use the project client alone, since it sends prompts to models directly
  2. Create an Azure Language client and pass the prompt as input text
  3. Call the embeddings endpoint and convert the vector back into text
  4. Use the project client to get an OpenAI-compatible client, then call the model
Show the answer and the explanation

Correct answer: D

The guidance is to use the project client for setup and configuration, then use the OpenAI-compatible client for calling models and running agents.

Source: official documentation

Question 7 Identify AI concepts and capabilities

A legal team reviews a plan to send customer questions to a GPT model sold by Azure in Microsoft Foundry. They ask what happens to the prompts and completions. According to the Foundry data, privacy, and security documentation, which statement is correct?

  1. They are shared with OpenAI in anonymized form so that it can improve its models.
  2. They are not available to OpenAI and are not used to train foundation models.
  3. They are used to retrain the foundation model unless the customer opts out.
  4. They are available to other customers who deploy the same model version.
Show the answer and the explanation

Correct answer: B

The documentation states that prompts and completions are not available to OpenAI or other providers of models sold by Azure, and are not used to train generative AI foundation models without the customer's permission or instruction.

Source: official documentation

Question 8 Implement AI solutions by using Microsoft Foundry

A small team wants its first single-agent solution in Foundry Agent Service: an assistant with instructions, a chosen model, and a file search tool. It does not want to write application code or manage any containers. Which agent type should it create?

  1. A hosted agent packaged as a container
  2. A prompt agent defined through configuration
  3. Agent logic assembled in code on each Responses API call
  4. A voice-based prompt agent using voice live
Show the answer and the explanation

Correct answer: B

Prompt agents are defined by instructions, a model, and tools, and Foundry runs them with no code or infrastructure to manage.

Source: official documentation

Question 9 Identify AI concepts and capabilities

A new developer asks how a GPT model deployed in Foundry is able to answer a question typed into the chat. Which description matches how the Microsoft documentation explains this behavior?

  1. The model looks up the answer in a database of stored questions and answers.
  2. The model produces the words most likely to follow the text it received.
  3. The model searches the internet and copies the most relevant passage it finds.
  4. The model runs a dedicated question-answering routine separate from text generation.
Show the answer and the explanation

Correct answer: B

GPT models attempt to produce the next series of words most likely to follow the previous text; an answer appears because it is the most likely kind of response to a question.

Source: official documentation

Question 10 Implement AI solutions by using Microsoft Foundry

A developer's console chat app uses the Chat Completions API. Each request sends only the system message and the user's newest question. Users report that the follow-up "And what is its population?" gets an answer unrelated to the city they just discussed. What is the most likely cause and fix?

  1. The max output tokens value is too small; increase it for follow-ups.
  2. The temperature is too low; raise it so the model infers the context.
  3. The model has no memory; send the prior user and assistant turns each time.
  4. The system message is too long; remove it so the question has more room.
Show the answer and the explanation

Correct answer: C

The model has no memory between calls, so the app must send an updated transcript with each question or the model loses the context of earlier turns.

Source: official documentation

Frequently asked questions

Are there free questions, and do I need an account?

Yes. 25 questions of this certification are free in study mode, across all its domains, without a subscription and without registration. The account is created when you subscribe, at checkout.

How does the subscription work, and can I cancel?

One monthly subscription covers all certifications and the timed practice exams. There is no minimum period. You cancel online, at any time, with the "Cancel your contract" function in My account; cancellation takes effect at the end of the month already paid, and no further payment is taken.

Where do the questions come from?

Every question is original, written from the vendor's public documentation. Each answer comes with an explanation and a link to the documentation page it relies on. InfraPrep is not affiliated with the vendors, and certification names belong to their owners.

In which language are the questions?

The questions are written in English, also on the French version of the site.

Is the catalog updated?

Yes, the catalog changes over time: new certifications and new questions are added.