Question 7
Domain 1: Fundamentals of AI and MLAn AI practitioner has trained a model on a training dataset. The model performs well on the training data. However, the model does not perform well on evaluation data. What is the MOST likely cause of this issue?
Correct answer: D
Explanation
The model is overfit because it performs well on the training data but poorly on evaluation data, which matches the definition of overfitting: the model learns the training set too closely and fails to generalize. AWS describes this as a model that fits training data well but has weak performance on new data, or "evaluation data."
Why each option is right or wrong
A. The model is underfit.
Underfitting means the model performs poorly on training data and evaluation data.
B. The model requires prompt engineering.
Prompt engineering adjusts prompts for foundation models; it does not fix a trained model's generalization gap.
C. The model is biased.
Bias is systematic unfairness or skew in predictions, not the train-versus-evaluation performance pattern.
D. The model is overfit.
AWS explicitly defines overfitting as a model that learns the training set too closely and therefore generalizes poorly to new, unseen data; see Amazon SageMaker Developer Guide, “Model Fit: Underfitting versus Overfitting” and AWS “What Is Overfitting in Machine Learning?”[^401034][^101018]. The fact pattern matches that definition exactly: strong performance on the training dataset, but weak performance on evaluation data, which is the standard holdout used to measure generalization error rather than memorization of the training examples.