Techniques to Improve Generative AI Model Output →mediumMultiple Choice

Generative AI Leader Practice Question: Techniques to Improve Generative AI Model Output

A team is building a customer support assistant on Vertex AI using a foundation model. They notice the model occasionally invents policy details that don't exist in their internal documentation. They want the model to ground its answers in the company's knowledge base and provide citations. Which approach should they use?

⚠ Common exam trap

The trap here is assuming that fine-tuning or temperature adjustment can ground a model in a private, frequently updated knowledge base, when only retrieval at inference time provides that grounding.

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 Retrieval-Augmented Generation (RAG) with Vertex AI Search to retrieve relevant documents and include them in the prompt.

Retrieval-Augmented Generation retrieves relevant passages from an authoritative knowledge base and supplies them as context, so the model's answer is conditioned on real policy text rather than parametric memory. With Vertex AI Search, the system can also surface citations, letting users verify sources. This combination directly reduces fabricated policy details while preserving natural language fluency.

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

✗
Reduce the max output tokens to limit the length of the model's responses.
Why it's wrong here
Limiting output length may truncate responses but does nothing to ensure factual accuracy or grounding. The model can still hallucinate within a shorter response. This parameter controls verbosity, not correctness, and does not connect the model to the internal knowledge base or enable citations.
✗
Increase the temperature parameter to encourage more creative responses.
Why it's wrong here
Raising temperature increases randomness and diversity in token selection, which would make the model more likely to hallucinate, not less. This directly contradicts the goal of grounding answers in factual documentation and providing accurate citations. Higher temperature is useful for creative tasks but harmful when factual precision and source attribution are required.
✗
Fine-tune the model on a dataset of general customer service conversations.
Why it's wrong here
Fine-tuning on generic customer service conversations teaches tone and format but does not inject the company's specific policy facts. The model may still invent details not present in the training data. Fine-tuning is not a reliable way to ground answers in a dynamic knowledge base and does not provide citations to source documents.
✓
Use Retrieval-Augmented Generation (RAG) with Vertex AI Search to retrieve relevant documents and include them in the prompt.
Why this is correct
RAG retrieves relevant passages from the company's knowledge base and injects them into the model's context, so the model bases its response on actual documentation. Vertex AI Search can also return citations pointing to the source documents. This directly addresses hallucination of policy details by grounding generation in verified internal content.

Go deeper

Generative AI Leader Practice Question: Techniques to Improve Generative AI Model Output

Why each option matters

Option-by-option breakdown

Related to this question