AI0-001 AI Infrastructure and Technologies Practice Question
A financial institution runs a credit-scoring model that must comply with internal governance requiring that every individual prediction be traceable to the input features that drove it, and that the explanation be produced at inference time for each applicant. The model is a complex gradient-boosted ensemble. Which approach best satisfies the requirement to generate a per-prediction explanation for each applicant?
⚠ Common exam trap
The trap here is treating global interpretability artifacts, such as feature importance or a surrogate tree, as if they explain individual predictions.
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
✓
Apply SHAP values to each individual prediction
Per-prediction traceability requires a local explanation method that decomposes an individual score into feature contributions. SHAP provides exactly that by attributing the difference between a prediction and a baseline to each input feature, with efficient exact computation available for tree ensembles. Global summaries such as feature importance or surrogate trees describe aggregate behavior and cannot explain a single applicant's decision.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Report feature importances from the trained ensemble
Why it's wrong here
Feature importances summarize how much each variable contributed across the entire training set. They are global statistics and say nothing about an individual applicant's outcome, so two applicants with opposite profiles would receive identical explanations. That cannot satisfy a policy requiring each prediction to be traceable to the features that drove it.
- ✓
Apply SHAP values to each individual prediction
Why this is correct
SHAP assigns each feature a contribution to a single prediction based on cooperative game theory, so every applicant receives an explanation showing which input values pushed the score up or down. It works with tree ensembles through efficient exact algorithms, and its additive guarantees make the per-prediction attribution auditable, matching the governance requirement precisely.
- ✗
Increase the number of boosting rounds to improve model stability
Why it's wrong here
Adding boosting rounds can change accuracy and convergence but produces no explanation artifacts whatsoever. It also tends to make the ensemble more complex and harder to interpret, working against the governance goal. More trees do not reveal which features influenced a given applicant's score, so this choice does not address the requirement.
- ✗
Train a global surrogate decision tree on the ensemble's outputs
Why it's wrong here
A global surrogate approximates the ensemble's overall behavior with a simpler model, giving a broad picture of feature influence. Because it is one model for all predictions, it cannot explain why a specific applicant received a specific score, and its fidelity to the ensemble varies by region of the input space. It fails the per-individual traceability the governance policy demands.
About these practice questions
One of 962 original AI0-001 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 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.