When someone says use Generalized Least Squares to fix the model what does that mean?
It is a different method of estimating the parameters of a regression model when a model exhibits heteroskedasticity or a high degree of multicollinearity.
It has different model assumptions than ordinary least squares.