Azure AI Fundamentals Practice Questions
Prepare for AI-901 with more than an answer.
Unlock the full exam and previous versions
- v1Microsoft Azure AI Fundamentals 150 questions Current
- AI-900Legacy Azure AI Fundamentals 249 questions Locked
- Exam fee
- $99 USD
- Level
- Fundamentals
- Valid for
- Does not expire (Fundamentals-level certifications do not require renewal)
Domains covered on the exam 2
- Identify AI concepts and capabilities42.5%
- Implement AI solutions by using Microsoft Foundry57.5%
- 1
A hospital is training a custom AI model to assist doctors with charting. Before uploading historical patient charts to the training environment, the data engineering team runs a script to redact all patient names, social security numbers, and home addresses.
Which principle of responsible AI is the team ensuring?
Show answer details
Correct answer: C
The principle of Privacy and Security requires that AI systems protect sensitive data and comply with privacy laws (like HIPAA or GDPR). Redacting Personally Identifiable Information (PII) before training a model is a direct application of this principle. Fairness deals with bias, while Accountability deals with human oversight.
- 2
You are building a voice-enabled virtual assistant for a drive-through restaurant. The system must listen to the customer ordering through the microphone, convert the audio into text, process the order, and then read the order confirmation back to the customer through a speaker.
Which TWO speech workload capabilities must you implement? (Select TWO)
Show answer details
Correct answer: A, C
Speech synthesis, also known as text-to-speech, is required to take the processed text (the order confirmation) and convert it into audible speech played through the speaker. Speech recognition handles the input.
Speech recognition, also known as speech-to-text, is required to listen to the customer's audio input through the microphone and convert it into text that the AI can process. Speech synthesis is required for the output.
- 3
An autonomous drone uses an AI model to navigate through forests. The engineering team has programmed a fail-safe: if the AI model's confidence in object detection drops below 60% due to heavy fog or darkness, the drone immediately halts forward movement and slowly lands.
Which responsible AI principle is primarily being addressed by this fail-safe mechanism?
Show answer details
Correct answer: C
Reliability and safety dictate that AI systems should perform consistently and safely, especially under unexpected conditions or uncertainty. By programming the drone to halt and land when confidence drops due to environmental factors, the team is ensuring the system fails safely rather than causing harm. Transparency refers to explaining how the model works.
- 4
A software development team is building an AI-powered application that helps users navigate public transit. During the design phase, the team specifically ensures that the application provides voice-guided navigation and high-contrast visual modes to support users with visual impairments. Which Microsoft principle of responsible AI is the team actively demonstrating?
Show answer details
Correct answer: B
The principle of Inclusiveness dictates that AI systems should empower everyone and engage people across the full spectrum of human diversity. By specifically designing voice-guided navigation and high-contrast modes for visually impaired users, the team is ensuring the application is accessible and inclusive. Fairness relates to treating all groups equitably without bias, while accountability ensures humans maintain oversight of the AI system.
- 5
A financial institution deploys an AI chatbot to help customers choose investment products. The bank's compliance department requires that the chatbot explicitly states to users that it is an AI, explains how it arrives at its recommendations, and provides a link to documentation detailing its limitations. Which responsible AI principle is driving these compliance requirements?
Show answer details
Correct answer: A
Transparency ensures that AI systems are understandable to users. It requires developers to be clear about the fact that users are interacting with an AI, explain how the system works, and communicate its capabilities and limitations openly. The compliance department's requirements directly mandate transparency. Reliability and safety focus on consistent performance, while privacy and security focus on data protection.
- 6
Contoso Medical operates a nationwide network of clinics. They have recently implemented an AI-driven diagnostic triage system built using Microsoft Foundry.
The system analyzes patient intake forms and historical electronic health records (EHR) to assign a triage priority score (1-5). During the first month of pilot testing, the clinical governance board discovers that the AI system consistently assigns lower triage priority to elderly patients with specific chronic conditions compared to younger patients presenting with identical acute symptoms.
The board mandates that an internal review board must audit the model's decision-making process weekly and that a senior physician must manually approve any triage score of 4 or 5 before the patient is asked to wait.
Which two Microsoft responsible AI principles are primarily being addressed by the board's discoveries and subsequent mandates?
Show answer details
Correct answer: D
The scenario highlights two distinct principles. First, the AI treating elderly patients differently than younger patients with identical symptoms is a violation of Fairness (equitable treatment across demographics). Second, the board's mandate that a senior physician must manually approve high-risk triage scores introduces human oversight, which is the core of Accountability. The human is ultimately responsible for the AI's impact.
flowchart LR A[Patient Intake] --> B[AI Model] B -->|Identifies Demographic Bias| C[Fairness Concern] B -->|Generates Score| D{Score >= 4?} D -->|Yes| E[Physician Review] E --> F[Accountability Maintained] D -->|No| G[Standard Queue]
