your function is complex. But a complex variable is really TWO variables, bundled into one. In ths case, it appears the imaginary part of g is virtually constant to within floating point trash.
But that imaginary part is non zero. So it makes no sense to try to plot a complex variable using surf.
At best, you can plot the real and imaginary parts separately. Since the imaginary part is boring...
Well, the real part is also pretty darn boring. Only along one edge of the surface does anything happen.
And what did happen was not much. Still down in the least significant bits.
Your function is essentially constant to within an ability to compute it in double precision.