mediumMultiple Choice
Generative AI Leader Practice Question: Using a fine-tuned generative model to create…
A company is using a fine-tuned generative model to create marketing copy. They want to ensure that when the model references statistics, it provides citations to original sources. Which technique should they use?
⚠ Common exam trap
A common misconception is that fine-tuning alone can enforce citation behavior. Google recommends using grounding (e.g., via Vertex AI grounding or citation features) to ensure generated content references original sources.
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
✓
Grounding by referencing source documents in the output
Grounding by referencing source documents in the output directly ties the generated statistics to verifiable original sources, ensuring citation accuracy. This technique retrieves and cites specific passages from trusted documents, which is essential for responsible AI in marketing copy where factual claims must be traceable.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Model tuning with citation data
Why it's wrong here
Tuning on citation data teaches the format of citing, not retrieval of the actual source statistics, so the model can fabricate plausible-looking references. Tuning suits adapting tone or domain vocabulary; grounding citations requires retrieval-augmented generation, which fetches and attributes real sources at inference time.
- ✓
Grounding by referencing source documents in the output
Why this is correct
Grounding retrieves source documents and constrains generation to them, so statistics can be attributed to their originals rather than hallucinated. Referencing those documents in the output satisfies the requirement for verifiable citations, which fine-tuning alone cannot guarantee.
- ✗
Chain-of-thought prompting
Why it's wrong here
Chain-of-thought prompting elicits intermediate reasoning steps, not source attribution; the model can reason fluently to a fabricated statistic. It suits arithmetic or multi-step logic tasks. Citation requires retrieval-augmented generation, which retrieves source documents and returns them as references alongside the generated text.
- ✗
Confidence indicators
Why it's wrong here
Confidence indicators expose the model's internal certainty, which is often poorly calibrated and unrelated to whether a statistic traces to a real source. They suit triaging outputs for human review. Citations require retrieval-augmented generation, which grounds each statistic in a retrieved, attributable document.
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.