In multiple regression, the reason we adjust R-squared is because...

Regular R-squared grows with the number of independent variables, even if these do not provide additional explanatory power

No real reason. Both regular and adjusted R-squared are equally valid in this context

Sometimes a regular R-squared can be negative

It makes it easier to compare models using this statistic, compared to regular R-squared