NCA-GENL Trustworthy AI Practice Question
A global bank must demonstrate to regulators that its LLM-based loan-advisory chatbot treats applicants from different regions and demographic groups equitably. The compliance team asks for an evaluation approach that quantifies outcome disparities across protected groups and produces evidence suitable for audit. Which approach best meets this requirement?
⚠ Common exam trap
The trap here is accepting upstream documentation or overall accuracy as proof of fairness, when regulators require measured subgroup outcomes from the actual deployed system.
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
✓
Run structured bias and fairness evaluations that measure outcome metrics across defined demographic subgroups and retain the results as audit artifacts
Regulatory proof of equitable treatment requires measured outcomes per protected group, documented and retained. Structured fairness evaluation supplies subgroup disparity metrics tied to the deployed chatbot, creating repeatable audit evidence. Vendor model cards, aggregate satisfaction scores, and larger datasets describe inputs or overall performance but cannot show whether specific groups receive different advisory outcomes, which is what the compliance team must demonstrate.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Rely on the vendor's model card, which states the base model was trained on diverse data
Why it's wrong here
A model card describes aggregate training characteristics but does not measure how this bank's chatbot behaves for this bank's applicants across regions and groups. It offers no outcome metrics tied to the deployed system, so it cannot serve as audit evidence of equitable treatment. Regulators require evidence about the specific application, not generic upstream documentation.
- ✓
Run structured bias and fairness evaluations that measure outcome metrics across defined demographic subgroups and retain the results as audit artifacts
Why this is correct
Structured fairness evaluation computes quantitative disparity metrics, such as selection or approval rates, for each protected subgroup and documents the methodology and results. This produces exactly the repeatable, reviewable evidence regulators expect, and running it on the deployed chatbot captures disparities in real advisory outcomes rather than only in training data.
- ✗
Increase the size of the fine-tuning dataset to include more loan applications
Why it's wrong here
More data may improve overall accuracy but does not by itself remove or reveal group-level disparities, and it provides no measurement artifact for auditors. Without targeted fairness evaluation, scaling the dataset could even entrench existing historical biases present in past lending decisions. This approach changes training inputs rather than producing the required evidence of equitable outcomes.
- ✗
Monitor average user satisfaction scores across the entire customer base
Why it's wrong here
Aggregate satisfaction can look healthy while specific protected groups receive worse advice, because averaging hides subgroup disparities. It also measures perceived experience rather than equitable advisory outcomes, and it lacks the demographic breakdown regulators need. Without subgroup-level fairness metrics, this monitoring cannot demonstrate equitable treatment.
About these practice questions
One of 367 original NCA-GENL 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 NVIDIA exam blueprint
This NCA-GENL practice question is part of Courseiva's free NVIDIA 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 NCA-GENL exam.