AIF-C01 Guidelines for Responsible AI Practice Question
A company uses an AI system to automate loan approvals. The model uses demographic features and achieves high accuracy, but the company wants to ensure compliance with responsible AI guidelines. Which practice best balances performance and fairness?
⚠ Common exam trap
The AIF-C01 exam often tests the misconception that simply removing sensitive attributes from the dataset guarantees fairness, without considering proxy bias or the need for ongoing monitoring.
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
✓
Remove sensitive attributes and monitor for proxy bias
Removing sensitive attributes (e.g., race, gender) from the training data directly addresses fairness by preventing the model from explicitly using these features. However, simply removing them is insufficient; monitoring for proxy bias (e.g., zip code or income correlating with race) is critical to ensure the model does not inadvertently learn discriminatory patterns through correlated features. This approach balances performance by retaining predictive power from non-sensitive features while actively auditing for fairness violations.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Use demographic features but with minimal monitoring
Why it's wrong here
Retaining demographic features without monitoring leaves bias undetected, breaching responsible AI principles; fairness cannot be verified. It is tempting because demographic data can raise predictive accuracy, and it would suit scenarios where disparate impact is legally permitted and continuous bias auditing is in place.
- ✗
Use a complex black-box model and rely on post-hoc explanations
Why it's wrong here
Post-hoc explanations do not remove bias encoded in black-box weights, so fairness compliance remains unverifiable. It appeals because black-box models often achieve higher raw accuracy, and it would be correct where explainability is the only regulatory demand and fairness metrics are already satisfied.
- ✓
Remove sensitive attributes and monitor for proxy bias
Why this is correct
Removing sensitive attributes directly addresses the fairness constraint, while monitoring for proxy bias catches indirect discrimination that demographic features create through correlated variables. This preserves predictive performance better than suppressing the model entirely, satisfying the stem's requirement to balance accuracy against responsible AI compliance.
- ✗
Optimize the model solely for accuracy on historical data
Why it's wrong here
Optimising solely for accuracy on historical data entrenches the demographic bias already present in past lending decisions, so the model cannot satisfy the fairness requirement. Accuracy-only tuning suits scenarios where historical data is unbiased and no protected attributes influence outcomes, such as forecasting equipment failure from sensor readings.
Go deeper
Related to this question
About these practice questions
One of 862 original AIF-C01 practice questions on Courseiva, each with a full explanation and wrong-answer analysis — not exam dumps or protected exam content. Learn why practice questions differ from exam dumps →
JA
Written by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
This AIF-C01 practice question is part of Courseiva's free Amazon Web Services 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 AIF-C01 exam.