AI0-001 Machine Learning and Deep Learning Practice Question
A team trains a model to predict whether loan applicants will default. On the holdout set the model achieves 0.86 AUC, but when audited, applicants over 60 receive systematically higher risk scores than equally qualified younger applicants. The team must reduce this disparity while preserving predictive performance. Which action should they take first?
⚠ Common exam trap
The trap here is assuming that deleting the protected attribute eliminates bias, when correlated proxy features often preserve the disparity and can even mask it from casual inspection.
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
✓
Measure disparity across protected groups and inspect whether age-correlated features drive the scores.
Bias mitigation should begin with measurement, because the source of disparity determines the right remedy. Group-level fairness metrics and feature attribution reveal whether age itself, a correlated proxy, or biased labels drive the scores. Only after diagnosing the cause can the team choose an appropriate intervention, such as reweighting, proxy removal, or post-processing.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Remove the age column from the training data and retrain the model.
Why it's wrong here
Dropping age does not remove its influence because correlated features such as employment history or credit age can act as proxies. The model may still produce disparate scores through those proxies, and removing a feature also sacrifices legitimate predictive signal. Blind feature removal without diagnosis is a common but ineffective fairness fix.
- ✗
Increase model complexity so it can learn finer distinctions between applicants.
Why it's wrong here
A more complex model tends to fit whatever patterns exist in the training data, including historical bias, which can amplify the disparity rather than reduce it. Complexity does not address fairness directly and may worsen the problem while also increasing overfitting risk. Fairness requires deliberate measurement and mitigation, not more capacity.
- ✗
Lower the overall decision threshold to approve more applicants.
Why it's wrong here
Changing the global threshold shifts approval rates for everyone and does not target the specific disparity between age groups. It can even widen the gap if the model's scores are systematically higher for older applicants. This action also fails to identify why the disparity exists, so the underlying cause remains.
- ✓
Measure disparity across protected groups and inspect whether age-correlated features drive the scores.
Why this is correct
Before mitigating bias, the team must quantify it and locate its source. Computing group-level metrics such as demographic parity or equal opportunity, and examining feature importance for age-correlated variables, reveals whether the disparity comes from a proxy feature, label bias, or sampling. This diagnostic step is the prerequisite for any effective, targeted fairness intervention.
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.