Fundamentals of Generative AI →mediumMultiple Choice

Generative AI Leader Fundamentals of Generative AI Practice Question

A marketing team at a retail company is using a generative AI model on Vertex AI to produce product descriptions from short bullet lists. They observe that the model's outputs are fluent but frequently invent specifications, such as claiming a jacket is waterproof when no such attribute was provided. The team wants to reduce these fabrications without retraining the model. Which approach should they take?

⚠ Common exam trap

The trap here is assuming that tuning decoding parameters like temperature or top-p will eliminate hallucinations, when those settings only change randomness and cannot supply missing factual 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

✓

Ground the model's responses by providing the bullet points as context in the prompt and instructing it to use only that information.

Grounding the generation in the provided bullet points by including them as context and instructing the model to use only that information is the most direct way to prevent invented specifications. It leverages the source data already available at inference time and requires no retraining. Parameters such as temperature, top-p, and output length affect style and randomness, not factual fidelity to a given source.

Answer analysis

Option-by-option breakdown

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

✗
Lower the model's top-p value to 0.1 to force deterministic outputs.
Why it's wrong here
Reducing top-p narrows the sampling pool and makes outputs more predictable, but it does not supply the model with the missing product facts. A deterministic model can still confidently state an incorrect specification if that pattern is likely from its training data. Top-p tuning controls randomness, not factual grounding, so it fails to address the root cause of the invented attributes.
✗
Increase the model's temperature parameter to encourage more diverse outputs.
Why it's wrong here
Raising temperature increases randomness in token selection, which makes the model more likely to produce varied and creative text, but it also increases the chance of hallucination. For a scenario where the goal is to stop invented specifications, higher temperature works against the objective. It does not ground outputs in the provided bullet points or reduce fabrication.
✗
Increase the maximum output tokens to give the model more room to explain its reasoning.
Why it's wrong here
Extending the output token limit allows longer responses, which can actually give the model more opportunity to elaborate and introduce additional unsupported claims. It does not provide the model with the authoritative product data or instruct it to avoid speculation. Longer outputs do not improve factual accuracy and may worsen the hallucination problem in this scenario.
✓
Ground the model's responses by providing the bullet points as context in the prompt and instructing it to use only that information.
Why this is correct
Grounding the model with the supplied bullet points as explicit context, combined with an instruction to rely solely on that information, constrains generation to the provided facts. This directly reduces fabricated attributes without retraining. It is the standard prompt-level technique for improving factual consistency in Vertex AI generative AI applications when the source data is available at inference time.

Go deeper

Generative AI Leader Fundamentals of Generative AI Practice Question

Why each option matters

Option-by-option breakdown

Related to this question