“`html id=”5j9nqk”
AI-901

Microsoft Azure AI Fundamentals

Free AI-901 Certification Practice Exam

Prepare for the Microsoft Azure AI Fundamentals (AI-901) certification exam with this free practice exam. Test your knowledge of Artificial Intelligence and Azure AI services with exam-style questions and receive instant results.

About This Practice Exam

This free AI-901 Certification Practice Exam is designed to help candidates assess their readiness for the Microsoft Azure AI Fundamentals certification. The exam covers core AI concepts, machine learning fundamentals, computer vision, natural language processing, and generative AI services in Azure.

Skills Measured

✅ Describe Artificial Intelligence Workloads and Considerations
✅ Describe Fundamental Principles of Machine Learning
✅ Describe Features of Computer Vision Workloads
✅ Describe Features of Natural Language Processing Workloads
✅ Describe Features of Generative AI Workloads

Practice Exam Features

✅ Exam-Style Questions
✅ Instant Results
✅ Free Access
✅ Mobile Friendly
✅ Certification-Focused Content

Start Your Free Practice Exam

Good luck with your AI-901 preparation.

0%
0 votes, 0 avg
0

Report a question

You cannot submit an empty report. Please add some details.

Exam Instructions

  • Read each question carefully.
  • Select the best answer.
  • You may review previous questions before submission.
  • Use the 📌 bookmark icon beside the question number to mark difficult questions for review.
  • Detailed explanations are available after answering each question, while your final score will be displayed at the end of the exam.
  • The quiz will automatically submit when the timer expires.

Tip: Mark questions you're unsure about and review them before finishing the exam.

Good luck!

Time is up.

Your quiz has been submitted automatically.


A1-901 Free Certification Practice Exam

Get exam-ready for Microsoft Azure AI Fundamentals (AI-901). Practice with certification-focused questions, identify knowledge gaps, and receive instant results to measure your readiness before taking the official certification exam.

📋 Before You Start the Practice Exam

To save your score and track your progress, please enter your
Name and Email Address before starting the exam.

  • ✅ Your exam results will be stored in your personal exam history.
  • ✅ Review previous attempts and monitor your improvement over time.
  • ✅ Certificates and score reports (when applicable) will be linked to your information.
Important:
Please use the same email address every time you take a practice exam.
This helps us maintain accurate exam history, scores, and performance records.

🚀 Enter your details below and click Next to begin the exam.

1 / 40

1. During text generation, a generative AI model usually predicts one ________ at a time based on the prompt and the text already generated.

2 / 40

2. In the Python quickstart, after analyzing an image with prebuilt-imageSearch, the code reads the ________ field to print the image description.

3 / 40

3. Which THREE pairs are correctly matched? (Select THREE.)

4 / 40

4. Which TWO situations are the strongest indicators of a fairness risk in an AI solution? (Select TWO.)

5 / 40

5. A team is building a custom analyzer for scanned expense forms. The solution must read image-based PDFs and return page location and confidence for extracted fields.

Which TWO configuration choices best fit this requirement? (Select TWO.)

6 / 40

6. An AI app should continue to behave safely when users submit unexpected or adversarial prompts. This is a reliability and safety consideration because the system must resist ________.

7 / 40

7. A team is building an AI solution that summarizes medical notes for clinicians. Which consideration is most important for reliability and safety?

8 / 40

8. Which TWO statements are true about how generative AI models work? (Select TWO.)

9 / 40

9. A model that produces a human-readable sentence describing an existing image is using ________. Which answer best completes the sentence?

10 / 40

10. A design team needs a model for prompt-based visual creation. Which TWO capabilities most directly indicate an image generation model? (Select TWO.)

11 / 40

11. Which THREE pairs are correctly matched? (Select THREE.)

12 / 40

12. For each of the following statements, determine whether the statement is correct.

- Statement 1: A deployed multimodal model can accept image input in addition to text.
- Statement 2: A public image URL can be used as input for image interpretation.
- Statement 3: Every multimodal model supports multiple images in the same chat turn.

Select the correct combination.

13 / 40

13. An AI team wants one practical step to improve fairness reviews before deployment.

Which step is best?

Select only one answer.

14 / 40

14. Which TWO statements describe image-generation model capabilities? (Select TWO.)

15 / 40

15. You need to send an image to a deployed multimodal model in Python. Which TWO input approaches are supported for the image? (Select TWO.)

16 / 40

16. You are building a lightweight app that listens in English and returns translated text in French.

translation_config = speechsdk.translation.SpeechTranslationConfig(
    subscription=speech_key,
    region=service_region
)
translation_config.speech_recognition_language = "en-US"
translation_config._______("fr")

17 / 40

17. An AI team decides to authenticate to Azure resources without storing secrets in application code. This decision primarily strengthens ________. Which answer best completes the sentence?

18 / 40

18. A training app must read lesson text aloud to learners. Which capability should the app use?

19 / 40

19. An AI app will analyze support tickets that may contain personal information. Which two actions best support privacy and security? (Select TWO.)

20 / 40

20. You are calling a deployed MAI image model to create a picture from a text prompt.

endpoint = os.environ["AZURE_ENDPOINT"]

url = f"{endpoint}/______"

Which code should replace the missing section?

21 / 40

21. In responsible AI, ________ means that people remain answerable for how an AI system is designed, deployed, and monitored.

Which answer best completes the sentence?

Select only one answer.

22 / 40

22. A team wants to alert operators when a safety score drops below the accepted threshold.

safety_score = result["safety_score"]
if safety_score < 0.90:
    ________
else:
    print("Safety check passed")

23 / 40

23. A school plans to release an AI tutor that processes student conversations. Before broad rollout, which review is most directly aligned to this responsible AI principle?

24 / 40

24. A model that has not been customized or fine-tuned for a specific use case is called a ________.

25 / 40

25. Which statement best describes audio_url in a multimodal chat request?

26 / 40

26. Protecting prompts, outputs, and personal data from unauthorized access is a ________ consideration in responsible AI. Which answer best completes the sentence?

27 / 40

27. For each of the following statements, determine whether the statement is correct.

Statement 1: Fairness concerns can appear when training data underrepresents part of the population.
Statement 2: Fairness and inclusiveness are identical terms in Microsoft's responsible AI principles.
Statement 3: Comparing model behavior across demographic groups can help identify fairness issues.

Select the correct combination.

28 / 40

28. Which THREE pairs are correctly matched? (Select THREE.)

29 / 40

29. A model selected for chat completions that can accept text and images as input is a ________. Which answer best completes the sentence?

30 / 40

30. You want a client call that sends both text and an image to a model.

image_url = "https://example.com/item.png"

response = client.complete(
messages=[
SystemMessage("You are a helpful assistant."),
UserMessage(content=[
TextContentItem(text="Describe this product photo."),
# Missing line
]),
],
temperature=0,
)

Which code should replace the missing section?

31 / 40

31. A web app must capture spoken microphone input and show live captions while the user is talking. Which Azure Speech capability should you add to the app?

32 / 40

32. To fine-tune pitch, speaking rate, or pronunciation for synthesized voice output, submit ________.

33 / 40

33. Which THREE pairs are correctly matched? (Select THREE.)

34 / 40

34. Which THREE pairs are correctly matched? (Select THREE.)

35 / 40

35. You need to identify people, organizations, and locations mentioned in support emails.

Which text analysis technique should you use?

36 / 40

36. A bank uses AI to recommend whether a loan application should be escalated for manual review. Which design choice best reflects accountability?

Select only one answer.

37 / 40

37. Even when an organization uses managed AI services, the organization remains ________ for how the AI solution is used and governed.

Which answer best completes the sentence?

Select only one answer.

38 / 40

38. A team deploys an AI model that helps prioritize insurance claims. Which practice best supports accountability in the solution?

Select only one answer.

39 / 40

39. For each of the following statements, determine whether the statement is correct.
Statement 1: recognize_once_async is intended for a single utterance.
Statement 2: start_continuous_recognition_async is the better choice for long-running multi-utterance input.
Statement 3: Voice Live API combines speech recognition, generative AI, and text-to-speech in one interface.
Select the correct combination.

40 / 40

40. You are preparing a request for an image-generation model in Foundry.

payload = {
"model": deployment_name,
________
"width": 1024,
"height": 1024
}

Which code should replace the missing section?

Your score is

The average score is 0%

“`