I agree. A perturbative approach is useful (even essential) for calculation at low energy. Loop, for instance "has gravitons" when one is explicitly studying low density, nearly flat, geometries. Low energy=low curvature, so that kind of approximation works.
I don't take issue with that. It's not a good way to picture reality when you are thinking about big bang cosmology. I take issue with someone calling the flat picture the real "territory" and the GR picture a mere "map".