Generative AI Leader Practice Question: Business Strategies for Generative AI Solutions
A university wants to deploy a generative AI study assistant across six faculties. Each faculty has its own curriculum documents and privacy rules, and the central IT team must prevent one faculty's content from appearing in another faculty's answers. Which architecture decision best satisfies this governance requirement?
⚠ Common exam trap
The trap here is trusting prompt instructions to enforce data boundaries, when retrieval isolation must be enforced in the architecture before the model ever sees the context.
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
✓
Create a separate grounded data store and retrieval configuration per faculty, and route each request only to its own faculty's store.
Separating grounded data stores per faculty enforces isolation at the retrieval layer, so answers can only draw on the requesting faculty's curriculum regardless of prompt wording. Each faculty keeps control of its own privacy rules, while central IT operates a single assistant platform. This structural separation is the only option that prevents leakage by design rather than by model compliance.
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 the model separately on each faculty's curriculum and expose a single endpoint that selects the adapter at random.
Why it's wrong here
Fine-tuning can embed faculty content in model weights, and random adapter selection would sometimes answer with the wrong faculty's knowledge, which is precisely the leakage the university must avoid. Fine-tuning also makes content removal difficult because the data is baked into parameters, complicating compliance when a faculty revises or withdraws material.
- ✗
Deploy the assistant in a single project with one service account and apply IAM conditions based on the user's email domain.
Why it's wrong here
IAM conditions can gate who may call the API, but they do not partition which documents retrieval may return, so a permitted user could still receive content from a faculty whose material they should not see. One shared service account also removes per-faculty attribution, weakening the audit posture the governance requirement implies.
- ✗
Use one shared data store for all curricula and rely on prompt instructions telling the model to answer only from the requesting faculty's material.
Why it's wrong here
A single shared index means retrieval can surface another faculty's documents before the prompt is even composed, and prompt instructions cannot reliably undo that exposure. Model behavior is probabilistic, so a confident but incorrect answer could still leak restricted content. Governance requirements demand structural isolation, not a request the model may ignore.
- ✓
Create a separate grounded data store and retrieval configuration per faculty, and route each request only to its own faculty's store.
Why this is correct
Isolating each faculty's curriculum in its own grounded data store means retrieval can only return documents from the requesting faculty, so cross-faculty leakage is prevented by architecture rather than by instruction. Each faculty can also apply its own privacy rules to its store, and central IT retains one platform to operate across all six.
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.