AIF-C01 Fundamentals of Generative AI Practice Question
A bank is using Amazon Bedrock to summarize customer support transcripts. The summaries often contain factual inaccuracies (hallucinations). Which approach is most effective for reducing hallucinations?
⚠ Common exam trap
AWS often tests the misconception that adjusting sampling parameters (top-p, temperature) or fine-tuning alone can fix hallucinations, when in fact these methods do not provide factual grounding and RAG is the standard industry approach for reducing factual inaccuracies in generative AI.
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 RAG by grounding summarization on retrieved transcripts
Retrieval-Augmented Generation (RAG) grounds the model's output on actual retrieved chunks of the customer support transcripts, providing factual context that reduces the likelihood of hallucination. By retrieving relevant transcript segments and feeding them as context to the LLM, the model generates summaries based on verified source material rather than relying solely on its parametric knowledge, which is the primary cause of factual inaccuracies.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Decrease the top-p to 0.1
Why it's wrong here
Lowering top-p to 0.1 restricts sampling to a narrow token set, which can increase repetition and truncate accurate content rather than grounding outputs in the transcript. It is tempting because nucleus sampling controls randomness, and it is useful for highly deterministic tasks, but it does not supply factual grounding.
- ✗
Increase the model's temperature to make outputs more diverse
Why it's wrong here
Raising temperature increases sampling randomness, which amplifies hallucination rather than reducing it. It is tempting because higher temperature suits creative or brainstorming tasks where diverse wording is wanted, but summarisation grounded in transcripts requires low-randomness, deterministic decoding.
- ✗
Fine-tune a smaller model on a large dataset of transcripts
Why it's wrong here
Fine-tuning teaches style and format, not factual grounding, so it cannot stop the model inventing details absent from transcripts. It is tempting because fine-tuning is the right choice when you need consistent output structure or domain-specific tone, but hallucination reduction requires retrieval-augmented generation.
- ✓
Implement RAG by grounding summarization on retrieved transcripts
Why this is correct
Grounding summaries in retrieved transcript passages constrains generation to source text, directly countering the factual drift that causes hallucinations. Retrieval narrows the model's context to verified customer dialogue, so fabricated details lack support and are suppressed. This satisfies the stem's core requirement: reducing inaccuracies in Bedrock summarisation without retraining.
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.