AIF-C01 Fundamentals of AI and ML Practice Question
A company is using Amazon Comprehend for sentiment analysis on customer reviews. They notice that the sentiment is often incorrect for negative reviews with sarcasm. What is the likely cause?
⚠ Common exam trap
The AIF-C01 exam often tests the misconception that 'fine-tuning' or 'more data' can fix any NLP issue, but here the trap is that sarcasm is a distinct linguistic challenge that pre-trained models inherently fail at, regardless of domain or data volume, unless specifically addressed with sarcasm-aware training or custom classifiers.
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
✓
The pre-trained model cannot handle sarcasm well
Amazon Comprehend's pre-trained sentiment analysis models are trained on general text corpora and lack the ability to detect sarcasm, which relies on contextual cues, tone, and figurative language. Sarcasm often inverts the literal sentiment (e.g., 'Great job, as always' for a failure), and standard NLP models without explicit sarcasm detection or fine-tuning cannot reliably interpret this inversion. Therefore, the likely cause is that the pre-trained model cannot handle sarcasm well.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
The model is not fine-tuned for the domain
Why it's wrong here
Amazon Comprehend's pre-trained sentiment model is not customisable or fine-tunable by customers, so domain fine-tuning is not an available remedy. It is tempting because fine-tuning custom models for domain vocabulary is standard practise, and it would be correct for a bring-your-own-model workflow on SageMaker rather than a managed Comprehend API.
- ✓
The pre-trained model cannot handle sarcasm well
Why this is correct
Amazon Comprehend's pre-trained sentiment model learns patterns from literal text and lacks the contextual reasoning needed to detect sarcasm, where stated sentiment inverts intended meaning. Sarcastic negative reviews therefore get classified by surface wording, producing incorrect sentiment labels.
- ✗
Insufficient training data
Why it's wrong here
Amazon Comprehend's sentiment analysis is a fully managed pre-trained service; customers cannot supply training data, so volume of training data is not a lever they control. It is tempting because insufficient data genuinely degrades custom-trained models, and it would be correct if the company had trained its own sentiment classifier on SageMaker.
- ✗
The input text is too long
Why it's wrong here
Comprehend accepts documents up to 5,000 bytes and returns an error for oversized input rather than silently misclassifying sentiment, so length does not explain wrong labels. It is tempting because truncation does harm accuracy in some pipelines, and it would be correct if the reviews were being cut off before reaching the model.
Go deeper
Related to this question
About these practice questions
One of 862 original AIF-C01 practice questions on Courseiva, each with a full explanation and wrong-answer analysis — not exam dumps or protected exam content. 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.