Generative AI Leader Google Cloud's Generative AI Offerings Practice Question
A bank is building a Gemini-powered assistant on Vertex AI that must answer questions about internal policy documents and must not fabricate policy details. The architects want to reduce hallucination and provide auditable sourcing. Which two Google Cloud capabilities should they combine? (Choose two.)
⚠ Common exam trap
The trap here is treating fine-tuning as a substitute for grounding, when fine-tuning shapes style and behavior but does not retrieve or cite the current authoritative documents.
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
✓
Enabling citations that map generated statements to retrieved source chunks
Grounding retrieves authoritative passages so responses are conditioned on real policy text, and citations expose exactly which passages supported each statement. Together they reduce fabrication and create an auditable chain from answer to source. Generation settings and safety controls do not supply factual anchoring, and fine-tuning on unrelated marketing material cannot substitute for retrieval over the actual policy corpus.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✓
Enabling citations that map generated statements to retrieved source chunks
Why this is correct
Citation support ties each generated claim back to the specific retrieved chunk it came from, giving reviewers a verifiable trail. Combined with grounding, this satisfies the bank's need for auditable sourcing and lets compliance staff confirm that a stated policy actually exists in the corpus, which is central to the stated requirement.
- ✗
Fine-tuning the model on the bank's public marketing brochures
Why it's wrong here
Fine-tuning on marketing brochures teaches tone and style from promotional content, not authoritative policy text, and it does not provide retrieval or citations at inference time. It cannot guarantee that answers reflect current internal policy, and stale or off-domain training data could even reinforce incorrect statements, so it does not meet the requirement.
- ✗
Disabling safety filters to allow unrestricted policy answers
Why it's wrong here
Safety filters govern harmful-content categories and are unrelated to factual grounding. Turning them off does not make answers more accurate and could allow inappropriate output, increasing risk without addressing hallucination. It fails to contribute to either the accuracy or the auditability goal and weakens overall governance.
- ✗
Setting the model temperature to its maximum value
Why it's wrong here
Raising temperature increases randomness in token selection, which makes outputs more varied and less deterministic. For a policy assistant where accuracy and reproducibility matter, high temperature would increase the chance of drifting from source content and is therefore counterproductive to reducing hallucination in this regulated scenario.
- ✓
Grounding with Vertex AI Search over an indexed policy data store
Why this is correct
Grounding constrains the model's response to retrieved passages from the indexed policy corpus, which directly reduces fabrication because the model is conditioned on real documents rather than relying only on parametric memory. It also yields citations back to the source passages, supporting the auditability the bank requires, making it a necessary component of the solution.
Go deeper
Related to this question
About these practice questions
Courseiva writes every Generative AI Leader question from scratch — 1,008 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 Google Cloud exam blueprint
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.