AI0-001 AI Security, Ethics and Governance Practice Question
A healthcare analytics company has trained an AI model to predict patient readmission risk using a dataset that includes ZIP code, race, and historical healthcare costs. Before deployment, the compliance team runs a fairness audit and finds that the model's predictions correlate strongly with race even though race was not used as a direct input feature. Which of the following BEST describes this phenomenon?
⚠ Common exam trap
The trap here is assuming that removing the protected attribute from the feature set makes a model fair, when correlated proxy variables can preserve the same bias.
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
✓
Proxy discrimination, where a seemingly neutral feature acts as a substitute for a protected attribute.
The model never receives race directly, but ZIP code and historical cost encode race-related disparities from decades of unequal healthcare access and residential segregation. This is proxy discrimination, and it defeats naive fairness approaches that only drop the protected column. Effective mitigation requires measuring disparate impact and auditing correlated features, because removing the explicit attribute does not remove its statistical footprint.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Overfitting, because the model memorized race-related patterns in the training set instead of generalizing.
Why it's wrong here
Overfitting describes poor generalization to new data, typically showing a large gap between training and validation performance. Here the model's race correlation is a systematic learned relationship caused by correlated input features, not memorization noise. The scenario gives no evidence of a train-validation performance gap, so overfitting is not the best explanation.
- ✗
Label leakage, because the target variable inadvertently encodes race information.
Why it's wrong here
Label leakage occurs when information about the target is unintentionally included in the features, inflating performance. Here the issue is that input features correlate with a protected attribute, producing biased predictions, not that the label contains future or hidden information. The readmission label itself is the outcome being predicted, so leakage does not describe this fairness problem.
- ✓
Proxy discrimination, where a seemingly neutral feature acts as a substitute for a protected attribute.
Why this is correct
ZIP code and historical cost are correlated with race due to historical segregation and unequal access to care, so the model learns race-associated patterns indirectly. This is proxy discrimination: a protected attribute is not an explicit input, yet its influence enters through correlated features. Detecting it requires disparate impact testing and feature-correlation analysis, not just removing the protected column.
- ✗
Data drift, because patient demographics changed after the model was trained.
Why it's wrong here
Data drift refers to changes in input data distribution over time after deployment, degrading model performance. This scenario describes a fairness audit conducted before deployment on the training data itself, so temporal distribution shift has not occurred. The correlation with race comes from the dataset's feature relationships, not from changing patient populations.
About these practice questions
Courseiva writes every AI0-001 question from scratch — 962 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 and reviewed by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
Last reviewed September 2026 · checked against the official CompTIA exam blueprint
This AI0-001 practice question is part of Courseiva's free CompTIA 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 AI0-001 exam.