Underfitting is the opposite problem: the model didn't study enough and never really understood the pattern.
Imagine it like this: A student who barely opened their textbook and only remembered a single very basic fact will do poorly both on practice problems and on the real exam — because they never really grasped the pattern in the first place.
Examples:
So? Underfitting means the model is too simple or didn't get enough good examples — the fix is usually better features or more training.