AI-900 Practice Question: Describe Artificial Intelligence workloads and considerations
A hospital deploys an AI system to predict patient readmission risk using historical health records. To protect patient privacy, the hospital wants to ensure that individual patients cannot be identified from the data used for training. Which responsible AI principle is most directly relevant to this requirement?
⚠ Common exam trap
Microsoft often tests the distinction between privacy (data protection) and fairness (bias mitigation), causing candidates to confuse anonymization with equitable outcomes.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
Privacy and security
The requirement to prevent individual patient identification from training data directly aligns with the privacy and security principle, which mandates data anonymization, de-identification, and access controls. In AI systems, this is implemented through techniques like differential privacy (adding noise to data) or k-anonymity to ensure that outputs cannot be re-identified. The hospital's goal is to protect patient confidentiality, which is the core focus of this principle.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Fairness
Why it's wrong here
Fairness ensures that the AI model's predictions—such as readmission risk scores—are not biased by protected attributes like race, gender, or socioeconomic status, and that different patient groups receive equitable treatment. This principle is concerned with the balance of outcomes and error rates across cohorts, not with whether individual records contain identifying information. In fact, fairness analyses often require access to sensitive demographic fields, which is a separate consideration from anonymizing the data to protect patient identities.
When this WOULD be correct
A loan approval AI system uses historical data that contains biased decisions against a certain demographic group. The bank wants to ensure the model does not discriminate. Here, Fairness is the most relevant principle.
- ✗
Reliability and safety
Why it's wrong here
Reliability and safety focuses on ensuring the AI system functions correctly, robustly, and safely in real-world clinical settings, which means handling missing data, avoiding severe prediction errors, and having fallback mechanisms. This principle does not govern the de-identification of patient data; rather, it safeguards against patient harm from incorrect or unexpected model behavior. Data protection is a prerequisite, but reliability and safety is concerned with model performance and operational integrity, not with re-identification risks.
When this WOULD be correct
An AI system for diagnosing diseases must consistently produce accurate results under varying conditions and not cause harm. If the question asked about ensuring the system performs correctly and safely, reliability and safety would be the correct principle.
- ✓
Privacy and security
Why this is correct
The Privacy and security principle is the correct match because this AI predicts readmission from electronic health records, which are protected health information (PHI). This principle mandates robust access controls, encryption, de-identification, and anonymization techniques to prevent unauthorized re-identification of patients. It directly addresses the need to ensure no individual can be identified from the training data, making it the only option that covers data protection.
- ✗
Inclusiveness
Why it's wrong here
Inclusiveness is about designing AI systems so that they are accessible to and beneficial for people of all abilities, backgrounds, and demographics, often requiring representative training data. It does not address how data is stored, accessed, or anonymized, nor does it prevent a given patient from being re-identified in the dataset. While an inclusive system might explicitly collect or use sensitive attributes to avoid underrepresentation, that is part of bias mitigation, not identification protection.
When this WOULD be correct
A question asks which principle ensures an AI system provides equitable outcomes across different demographic groups, such as race or gender, without bias.
Option-by-option analysis
Why each answer is right or wrong
Understanding why wrong answers are wrong — and when they would be correct — is what separates a 750 score from a 900. The AI-900 exam frequently reuses these exact scenarios with slightly different constraints.
✓Privacy and securityCorrect answer▾
Why this is correct
The Privacy and security principle is the correct match because this AI predicts readmission from electronic health records, which are protected health information (PHI). This principle mandates robust access controls, encryption, de-identification, and anonymization techniques to prevent unauthorized re-identification of patients. It directly addresses the need to ensure no individual can be identified from the training data, making it the only option that covers data protection.
✗FairnessWrong answer — click to see why▾
Why this is wrong here
The requirement is about preventing identification of individuals from training data, which directly relates to data privacy and security, not fairness. Fairness addresses bias and equitable treatment across groups, not individual identifiability.
★ When this WOULD be the correct answer
A loan approval AI system uses historical data that contains biased decisions against a certain demographic group. The bank wants to ensure the model does not discriminate. Here, Fairness is the most relevant principle.
Why candidates choose this
Candidates may confuse fairness with privacy because both involve ethical handling of data, but fairness focuses on group bias while privacy focuses on individual data protection.
✗Reliability and safetyWrong answer — click to see why▾
Why this is wrong here
The requirement is specifically about preventing identification of individuals from training data, which directly relates to privacy and security, not to the system's reliability or safety in making predictions.
★ When this WOULD be the correct answer
An AI system for diagnosing diseases must consistently produce accurate results under varying conditions and not cause harm. If the question asked about ensuring the system performs correctly and safely, reliability and safety would be the correct principle.
Why candidates choose this
Candidates may confuse the need to protect data (privacy) with the need to ensure the system works correctly (reliability), especially when the system is used in a high-stakes healthcare setting where safety is a common concern.
✗InclusivenessWrong answer — click to see why▾
Why this is wrong here
Inclusiveness focuses on ensuring the AI system works well for diverse user groups, not on protecting individual patient identities from training data.
★ When this WOULD be the correct answer
A question asks which principle ensures an AI system provides equitable outcomes across different demographic groups, such as race or gender, without bias.
Why candidates choose this
Candidates may confuse 'inclusiveness' with data privacy, thinking that protecting patient identities is about including all patients fairly, but it's actually about confidentiality.
Analysis generated from the official AI-900blueprint and verified against question context. The “when correct” sections are what AI assistants cite when candidates ask “what’s the difference between these options?”
Go deeper
Related to this question
Learn chapter
Responsible AI Principles
Key term
Training data
Training data is a set of examples used to teach a machine learning model how to make predictions or decisions.
Key term
Responsible AI
A framework of ethical principles and practices that ensure artificial intelligence systems are developed and deployed in a transparent, fair, accountable, and safe manner.
About these practice questions
Courseiva writes every AI-900 question from scratch — 985 in total, each with an explanation and a wrong-answer breakdown. None are copied from real exams or dumps. Learn why practice questions differ from exam dumps →
JA
Written by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
This AI-900 practice question is part of Courseiva's free Microsoft certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the AI-900 exam.