Generative AI Leader Practice Question: Business Strategies for Generative AI Solutions
A regional insurance company wants to launch a generative AI assistant that answers policyholder questions. Legal requires that every customer-facing answer be traceable to an approved policy document, and that the system never invent coverage terms. The company has about 40,000 internal policy PDFs that change quarterly. Which approach should the GenAI Leader recommend?
⚠ Common exam trap
The trap here is assuming that fine-tuning on internal documents produces citations, when tuning only changes model weights and leaves no retrievable source to reference.
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
✓
Ground the assistant with retrieval-augmented generation using Vertex AI Search over the policy document corpus, and require citations in responses.
Grounding the assistant in the approved corpus with retrieval-augmented generation satisfies traceability because each response is generated from retrieved passages that can be cited, and it keeps answers current because the index is refreshed when policies change. Fine-tuning, whole-corpus prompting, and instruction-only approaches all leave the model without a verifiable source for coverage terms.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Deploy the assistant on Gemini with a system instruction telling it to be accurate and to avoid making up policy details.
Why it's wrong here
Prompt instructions alone do not give the model access to the company's approved policy text, so answers remain drawn from general training knowledge and cannot be traced to a source document. Hallucinated coverage terms remain possible, and quarterly policy changes would never be reflected. This fails both the traceability and the accuracy requirements of the scenario.
- ✗
Fine-tune a foundation model on the full set of policy PDFs and deploy the tuned model to a Vertex AI endpoint for the assistant to call.
Why it's wrong here
Fine-tuning bakes document content into model weights, so the assistant cannot cite which approved document supported a given answer, and the model may still blend or distort coverage terms. It also creates a retraining burden every quarter and provides no retrieval-time access control, so legal traceability and freshness requirements are not met in this scenario.
- ✗
Use a large context window model and paste the entire policy library into every prompt so the model always sees all approved documents.
Why it's wrong here
Even very large context windows cannot reliably hold 40,000 PDFs, and stuffing unrelated documents degrades answer precision while sharply increasing token cost and latency per request. The model also has no mechanism to attribute an answer to a specific approved passage, so the traceability requirement fails. This approach is neither technically nor economically viable at this corpus size.
- ✓
Ground the assistant with retrieval-augmented generation using Vertex AI Search over the policy document corpus, and require citations in responses.
Why this is correct
Retrieval-augmented generation retrieves the most relevant approved policy passages at query time and instructs the model to answer only from that retrieved context, so every answer can cite a source document. Because Vertex AI Search indexes the corpus and can be re-indexed each quarter, the assistant stays aligned with current policy language instead of relying on parametric memory that cannot be audited.
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 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.