mediumMultiple Choice
AIF-C01 Practice Question: Developing an LLM-powered application that…
A company is developing an LLM-powered application that generates investment advice. They are concerned about the model producing inaccurate or fabricated information. Which combination of techniques should they implement to minimize hallucinations?
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 Retrieval-Augmented Generation (RAG) and use Bedrock Guardrails
RAG grounds the model in retrieved factual documents, and Bedrock Guardrails can filter or block content that contradicts known facts or is speculative. Prompt engineering alone is insufficient. Fine-tuning reduces but does not eliminate hallucinations. Reducing temperature may reduce creativity but doesn't ground the model.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Fine-tune the LLM on a dataset of correct investment advice
Why it's wrong here
Fine-tuning on correct advice teaches style and domain phrasing but does not ground outputs in verifiable sources, so fabrication persists. It is tempting because fine-tuning is a genuine adaptation technique, and it would be correct for instilling format or tone rather than preventing hallucination.
- ✗
Use a larger LLM and rely on its pre-trained knowledge
Why it's wrong here
Scaling up a pre-trained model does not ground answers in verified data; the weights still generate fluent fabrications when uncertain. Retrieval-augmented generation with citations, plus guardrails, is what constrains output to source documents. Larger models are chosen for reasoning depth or broader general knowledge, not factual reliability.
- ✓
Implement Retrieval-Augmented Generation (RAG) and use Bedrock Guardrails
Why this is correct
Retrieval-Augmented Generation grounds responses in retrieved, verifiable documents rather than relying solely on parametric memory, directly reducing fabricated investment claims. Bedrock Guardrails adds contextual grounding and automated reasoning checks that filter unsupported statements. Together they satisfy the stem's requirement to minimise hallucinations in generated advice.
- ✗
Use a lower temperature setting and increase the max token count
Why it's wrong here
Lower temperature reduces sampling randomness, but a larger max token count lengthens output and gives fabricated claims more room to accumulate. It is tempting because both are generation parameters, and lowering temperature alone would help if the goal were more deterministic, factual responses.
Go deeper
Related to this question
About these practice questions
Courseiva writes every AIF-C01 question from scratch — 862 in total, each with an explanation and a wrong-answer breakdown. None are copied from real exams or dumps. Learn why practice questions differ from exam dumps →
JA
Written by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
This AIF-C01 practice question is part of Courseiva's free Amazon Web Services 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 AIF-C01 exam.