AIF-C01 Practice Question: Deploying a large language model (LLM) for…

A company is deploying a large language model (LLM) for customer support. They want to reduce the risk of hallucinations. Which TWO approaches should they implement? (Choose two.)

⚠ Common exam trap

The trap is assuming that simply increasing model size or token limit will reduce hallucinations, when in fact these can exacerbate the problem. Candidates may also overlook that fine-tuning on bad examples worsens the issue.

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

✓

Implement Amazon Bedrock Guardrails to define topics the model should avoid

Option A is correct because Amazon Bedrock Guardrails lets you define denied topics, content filters, and contextual grounding checks that constrain what the model can say, directly reducing the chance it produces off-topic or fabricated responses in a customer support setting. Option E is correct because Retrieval-Augmented Generation (RAG) retrieves authoritative documents from a knowledge base and injects that factual context into the prompt, grounding the model's answers in real data rather than relying solely on parametric memory, which is the standard mitigation for hallucinations. Option B is wrong because a larger model without retrieval still generates from learned parameters and can hallucinate confidently; scale alone does not guarantee factual accuracy. Option C is wrong because fine-tuning on hallucinated examples would teach the model to reproduce fabrications, worsening the problem. Option D is wrong because increasing max token length only allows longer outputs and does not improve factual grounding; it can even give the model more room to elaborate on incorrect claims.

Answer analysis

Option-by-option breakdown

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

✓
Implement Amazon Bedrock Guardrails to define topics the model should avoid
Why this is correct
Guardrails apply configurable content filters and topic-deny policies that block the model from generating responses outside approved subject areas. This constrains outputs at inference time, reducing the chance of fabricated or off-topic answers, which addresses the stated hallucination-reduction requirement.
✗
Use a larger model without any retrieval mechanism
Why it's wrong here
Scaling parameters improves fluency and reasoning but does not ground outputs in verified sources, so fabrication persists. It is tempting because larger models hallucinate less on some benchmarks, yet retrieval augmentation, not parameter count, supplies the factual anchor this scenario requires.
✗
Fine-tune the model on a dataset of hallucinated examples
Why it's wrong here
Training on hallucinated examples teaches the model to reproduce fabrication, reinforcing the exact behaviour to eliminate. It is tempting because fine-tuning on domain data is a legitimate grounding technique, but the dataset content, not the method, determines whether hallucination decreases.
✗
Increase the maximum token length to give the model more room to elaborate
Why it's wrong here
A longer token budget lets the model generate more ungrounded text, increasing fabricated content rather than constraining it. It is tempting because token limits do govern output length, but grounding via retrieval or constrained decoding is what reduces hallucination.
✓
Use Retrieval-Augmented Generation (RAG) to provide factual context
Why this is correct
RAG grounds the model's responses in retrieved, authoritative documents rather than relying solely on parametric memory, directly reducing fabricated content. By supplying factual context at inference time, it constrains the model to cite real source material, satisfying the requirement to lower hallucination risk.

Go deeper

AIF-C01 Practice Question: Deploying a large language model (LLM) for…

Why each option matters

Option-by-option breakdown

Related to this question