mediumMultiple Choice
Generative AI Leader Practice Question: A machine learning engineer is evaluating a…
A machine learning engineer is evaluating a generative AI model for bias. They have a diverse test set covering gender, race, and age groups. Which metric would best indicate if the model's performance is systematically worse for certain demographic groups?
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
✓
Equalized odds across demographic groups
Equalized odds measures whether a model's predictions have equal false positive/negative rates across groups. The other options either measure different aspects or are not specific to fairness.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Model perplexity on held-out data
Why it's wrong here
Perplexity measures how well a language model predicts held-out text overall; it collapses all demographic groups into one aggregate score, so systematic underperformance for a subgroup stays hidden. It is tempting because perplexity is a standard generative-quality metric, but it would be the right choice for comparing language-modelling fluency, not bias.
- ✓
Equalized odds across demographic groups
Why this is correct
Equalised odds compares true positive and false positive rates across demographic groups, exposing performance gaps that aggregate accuracy hides. It directly satisfies the stem's need to detect systematically worse performance for specific gender, race, or age groups.
- ✗
Overall accuracy on the test set
Why it's wrong here
Overall accuracy averages performance across every demographic group, so strong results for a majority group can mask systematically worse outcomes for a minority one. It is tempting because accuracy is the default classification metric, and it would be correct when groups are balanced and aggregate performance is the only concern, not subgroup fairness.
- ✗
Area under the ROC curve (AUC)
Why it's wrong here
AUC measures overall ranking quality, not group-specific disparities.
Go deeper
Related to this question
About these practice questions
This Generative AI Leader question is part of Courseiva's 1,008-question bank — original exam-style content with full explanations and wrong-answer analysis, never real exam questions or exam dumps. Learn why practice questions differ from exam dumps →
JA
Written by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
This Generative AI Leader practice question is part of Courseiva's free Google Cloud 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 Generative AI Leader exam.