Adjusted $R^2$ always increases.

Adjusted $R^2$ can decrease if the added predictor does not sufficiently improve the model fit.

Adjusted $R^2$ and $R^2$ are always equal.

Adjusted $R^2$ is always greater than $R^2$.