Techniques to Improve Generative AI Model Output →mediumMultiple Select

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

A media company uses Gemini on Vertex AI to generate short news summaries from long articles. The summaries frequently miss key facts and sometimes include details not in the source. The team wants to improve factual grounding and coverage without retraining the model. (Choose two.)

⚠ Common exam trap

The trap here is assuming that infrastructure scaling or streaming features improve factual accuracy, when grounding requires better context and prompting rather than more compute.

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 supply relevant source passages in the prompt.

Retrieval-augmented generation supplies authoritative source passages at inference time, and few-shot examples teach the model the desired grounded summarization pattern. Together they improve fact coverage and reduce unsupported claims without retraining. Temperature tuning, larger machines, and streaming do not add source knowledge or enforce faithfulness, so they fail to solve the stated problem.

Answer analysis

Option-by-option breakdown

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

✓
Use retrieval-augmented generation (RAG) with Vertex AI Search to supply relevant source passages in the prompt.
Why this is correct
RAG retrieves authoritative passages from the article or an indexed corpus and injects them into the model context, so the summary is conditioned on real source text. This directly reduces hallucination and improves fact coverage without fine-tuning. It also lets you update the knowledge base independently of the model.
✗
Enable streaming responses so the model can revise earlier sentences as it generates.
Why it's wrong here
Streaming returns tokens incrementally to the client for perceived latency benefits, but it does not let the model revise already emitted tokens. It has no effect on factual grounding or coverage. This option misrepresents how autoregressive decoding works and would not address the summarization quality issue.
✗
Deploy the model on a larger machine type with more vCPUs to increase factual accuracy.
Why it's wrong here
Compute size affects throughput and latency, not the factual correctness of generated text. A larger machine type will not add source knowledge or reduce hallucination. Accuracy improvements come from prompt design, retrieval, or tuning, not from infrastructure sizing. This choice confuses performance scaling with quality.
✓
Add few-shot examples in the prompt that show correctly grounded summaries with citations to the source.
Why this is correct
Few-shot prompting demonstrates the desired output format and behavior, including how to cite source facts and avoid unsupported claims. The model learns the pattern from the examples at inference time, improving consistency and grounding without any training job. This is a low-cost prompt engineering technique.
✗
Increase the model temperature to 1.5 so the model explores more of the source content.
Why it's wrong here
Higher temperature increases randomness and creativity, which makes summarization less faithful to the source. It would amplify invented details rather than improve grounding. For factual summarization, lower temperature is generally preferred. This setting does not add missing source facts to the context.

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