Overfitting
When an AI memorises its practice examples instead of learning the general idea, so it fails on new ones.
In everyday terms
A model that scores 100% on its training data but poorly in the real world has overfit. It learned the noise, not the signal.
For professionals
High variance: training loss keeps falling while validation loss rises. Countered with more data, regularisation, early stopping and simpler models.
Think of it like…
A student who memorised last year's exam answers but can't handle a new question.
You've already seen it
An AI that is brilliant in a demo but disappointing on your own data.
Myth vs reality
Myth: Perfect accuracy on training data means a great model.
Reality: It often signals overfitting. What matters is accuracy on unseen data.
Quick check
A model aces its training set but fails on new data. This is…
- Underfitting
- Overfitting
- Inference
- Alignment
Show answer
Overfitting: It memorised instead of generalising.