Overfitting
When an AI model learns its training data too well and fails to generalise to new data.
Overfitting happens when a model memorises the training data instead of learning the underlying patterns. It performs extremely well on data it has seen before, but poorly on new, unseen examples. The model has essentially learned the noise along with the signal.
Imagine a student who prepares for an exam by memorising every past paper word for word instead of understanding the subject. They ace any question from those papers but completely fall apart when a slightly different question appears. They overfitted to the revision material.
More training data doesn't automatically prevent overfitting. The quality and diversity of data matters more than quantity. A model trained on a huge but narrow dataset can still overfit badly. Regularisation techniques and validation sets are the real safeguards.