Generative AI Leader Google Cloud's Generative AI Offerings Practice Question
A logistics company must answer employee questions using its internal policy manuals, which change weekly. The manuals are stored in a Cloud Storage bucket and must never be used to train a shared model. The company wants relevant, up-to-date answers with citations. Which Google Cloud approach should it use?
⚠ Common exam trap
The trap here is treating fine-tuning as the way to add new facts, when grounding with a search data store is the mechanism that keeps answers current and attributable.
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
✓
Use Vertex AI Search with the Cloud Storage bucket as a grounded data store
Grounding responses in a Vertex AI Search data store keeps answers tied to the current contents of the Cloud Storage bucket and provides citations to the source passages. Because the documents are indexed rather than used to train a shared model, the company retains control over sensitive policy text. Fine-tuning and prompt stuffing both fail on freshness, data handling, or scalability.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Fine-tune a Gemini model on the policy manuals and redeploy it weekly
Why it's wrong here
Fine-tuning bakes information into model weights, so weekly manual changes would require repeated retraining and redeployment. It also risks embedding sensitive policy text into a tuned model, which conflicts with the requirement that the manuals never train a shared model. Fine-tuning is better suited to teaching style or task behavior than to frequently changing factual knowledge.
- ✓
Use Vertex AI Search with the Cloud Storage bucket as a grounded data store
Why this is correct
Vertex AI Search can index documents from Cloud Storage and ground Gemini responses in that content, so answers stay current as the bucket changes. It returns citations to source passages, and the manuals remain in the company's own data store rather than being used to train a shared model. This directly satisfies the freshness, attribution, and data-handling requirements.
- ✗
Export the manuals into prompt text and paste them into every request
Why it's wrong here
Pasting entire manuals into each prompt is impractical because of context limits and cost, and it offers no citation mechanism or automatic refresh when documents change. It also creates operational risk if sensitive text is copied into logs. This approach does not scale to a large, frequently updated policy library.
- ✗
Increase the Gemini model's temperature setting to improve factual recall
Why it's wrong here
Temperature controls randomness in generation, not access to external documents. Raising it would make outputs more varied and less predictable, which is the opposite of what a policy question-answering system needs. It cannot supply citations or keep answers synchronized with weekly manual updates.
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 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.