Free AWS AIF-C01 Certified AI Practitioner Actual Exam Questions - Question 13 Discussion
The generated content sounds plausible and factual but is incorrect.
Which problem is the LLM having?
B imo. Hallucination fits better because the model is generating info that sounds legit but isn’t actually true. Overfitting (C) usually means memorizing training data too closely, which would cause different problems like poor generalization, not making stuff up. Data leakage (A) would be about the model accidentally training on test data, which isn’t hinted at here. Underfitting (D) means the model hasn’t learned enough patterns, leading to poor performance overall, but not necessarily plausible-sounding false facts. So hallucination is the classic issue where the LLM invents wrong details w
Not C, overfitting usually means the model just memorized training data and doesn’t generalize well, but here the errors are made up content, so B makes more sense as it’s about generating false info.
Maybe B. The question highlights that the content sounds plausible but is incorrect, which matches hallucination where the model "makes up" facts rather than training errors like overfitting or underfitting.
B imo. The key clue is the content sounding believable but being factually wrong, which matches hallucination exactly, unlike overfitting or underfitting that relate to training issues.
B, because the model invents details that aren’t based on real data.
It’s B. Hallucination happens when the model makes stuff up that sounds right but isn’t actually true. Seen this happen a lot with content generation.