Generative AI Leader Google Cloud's Generative AI Offerings Practice Question
An organization is building a generative AI application on Vertex AI. Which THREE actions should they take to ensure responsible AI practices?
⚠ Common exam trap
The trap here is that candidates might think disabling content filtering (Option A) improves performance, but the exam tests that responsible AI on Google Cloud's Vertex AI requires both automated filters and human oversight, as emphasized in Google's AI Principles.
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
✓
Implement human review for sensitive outputs
Option B is correct because implementing human review for sensitive outputs ensures that high-risk or ambiguous AI-generated content is validated by a person before it reaches users, which is a core responsible AI safeguard for generative applications on Vertex AI. Option C is correct because conducting fairness evaluation (for example, using Vertex AI's model evaluation tools to assess bias across demographic groups) helps detect and mitigate discriminatory or skewed model behavior, directly supporting responsible AI. Option D is correct because creating a safety policy and enforcing it via content filtering operationalizes responsible AI by defining prohibited content and using Vertex AI safety filters (such as configurable hate speech, harassment, and dangerous content thresholds) to block violations. Option A is not correct because disabling content filtering removes a key safety control and increases the risk of harmful outputs, which is contrary to responsible AI. Option E is not correct because using only Google's foundation models does not by itself guarantee responsible AI; responsibility depends on evaluation, policy enforcement, monitoring, and human oversight regardless of which models are used.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Disable content filtering
Why it's wrong here
Disabling content filtering removes the safety layer that blocks harmful outputs, directly undermining responsible AI. It is tempting because filtering can reject legitimate prompts and add latency, so teams disable it to improve response quality or speed in internal testing environments.
- ✓
Implement human review for sensitive outputs
Why this is correct
Human review places a person in the loop to inspect sensitive outputs before they reach users, catching harmful or inaccurate content that automated filters miss. This directly satisfies the responsible AI requirement for oversight and accountability in the generative AI application built on Vertex AI.
- ✓
Conduct fairness evaluation
Why this is correct
Fairness evaluation measures model outputs across demographic groups to detect bias and disparate performance. Running this assessment on Vertex AI satisfies the responsible AI requirement by identifying and mitigating discriminatory behaviour before deployment, ensuring equitable treatment across user populations.
- ✓
Create a safety policy and enforce via content filtering
Why this is correct
A safety policy defines prohibited content categories, and content filtering enforces those rules automatically on prompts and responses. This satisfies responsible AI by preventing harmful outputs at scale, giving the Vertex AI application a consistent, auditable guardrail mechanism.
- ✗
Use only Google's foundation models
Why it's wrong here
Restricting the application to Google's own foundation models does not by itself satisfy responsible AI; governance, evaluation and monitoring do. It is tempting because using vetted first-party models reduces supply-chain and provenance risk, which is a genuine consideration when third-party model transparency is limited.
Go deeper
Related to this question
About these practice questions
One of 1,008 original Generative AI Leader 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 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.