It's not like there's a huge difference. Causal inference is a special case of curve-fitting, where one chooses what variables should enter into the fit according to a causal graph.