About a name
In statistics, y is the truth — the value actually observed in the world. ŷ ("y-hat") is the estimate — what a model predicts the truth to be.
Every model, from a weather forecast to a large language model, produces Å·. Never y. The hat is a small typographic honesty marker: this number was made by a machine, and it might be wrong.
The distance between the two, ŷ − y, is called the residual — the machine's error. Most of the work of building systems that can be trusted is not making predictions. It is knowing the residual exists, measuring it, and deciding what happens where the model turns out to be wrong.
Modern AI is very good at producing ŷ that looks like y — fluent, confident, plausible. The hat is easy to forget. Forgetting it is how AI projects fail.
y-hat is named for the gap, because the gap is where the real work is.