AIF-C01 Fundamentals of Generative AI Practice Question
A media company wants a generative AI assistant that drafts scripts in a distinctive house style. The team has several thousand pages of approved scripts but no labeled input-output pairs, and they want to adapt an existing foundation model rather than train one from scratch. Which approach best matches their data and goal?
⚠ Common exam trap
The trap here is reaching for supervised fine-tuning by instinct, even though the scenario states there are no labeled input-output pairs, which rules it out.
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
✓
Continued pre-training on the unlabeled script corpus to adapt the model to the house style
Continued pre-training is designed for exactly this situation: a large body of unlabeled domain text and a desire to shift an existing foundation model toward that domain. Supervised fine-tuning and RLHF both need labels the team does not have, and few-shot prompting cannot absorb thousands of pages of stylistic signal into the model itself.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Reinforcement learning from human feedback using preference rankings of generated scripts
Why it's wrong here
RLHF requires people to compare and rank multiple model outputs so a reward model can be trained. That is a costly labeling pipeline the team has not built, and it optimizes toward human preference rather than absorbing an existing corpus. It does not address the core need of learning style from thousands of unlabeled pages.
- ✗
Supervised fine-tuning on prompt-completion pairs extracted from the scripts
Why it's wrong here
Supervised fine-tuning expects curated input-output pairs that teach the model to follow instructions or produce specific responses. The scenario explicitly states there are no labeled pairs, and manufacturing them from raw scripts would require substantial annotation effort. The technique is valid in general but does not match the data the team actually has.
- ✗
Few-shot prompting with three example scripts inserted into each request
Why it's wrong here
Few-shot prompting can nudge tone, but only a handful of examples fit in a prompt, so it cannot internalize patterns spread across thousands of pages. Each request also pays the token cost of the examples. It is a lightweight adaptation that leaves model weights unchanged, which is weaker than what the volume of available data supports.
- ✓
Continued pre-training on the unlabeled script corpus to adapt the model to the house style
Why this is correct
Continued pre-training consumes large volumes of unlabeled domain text and adjusts the model's weights toward that distribution, which is precisely how a model absorbs a distinctive vocabulary, tone, and structure. Because the scenario has thousands of pages and no paired labels, this is the only listed method that uses the data as-is while still changing model behavior beyond what prompting alone provides.
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 and reviewed by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
Last reviewed September 2026 · checked against the official Amazon Web Services exam blueprint
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.