What issue does overfitting introduce to machine learning models?

Overfitting occurs when a machine learning model learns not only the underlying patterns in the training data but also the noise and outliers within that data, leading to a highly complex model. This results in the model performing exceptionally well on the training dataset, sometimes achieving high accuracy or low error. However, this performance often does not hold up when the model is exposed to new, unseen data, as it fails to generalize beyond the specific instances it was trained on.

In contrast to overfitting, a model that generalizes well will perform consistently across both training data and new data. Thus, while overfitting leads to high performance on the training set, it ultimately compromises the model's ability to generalize, which is a critical aspect of effective machine learning.