It is understood that finding a "universal""the" universal functional is an NP-complete problem. However, progressively better functionals have been constructed decade after decade and fitted to more and more comprehensive datasets.
What is the closest humans have achieved, to a universal functional?
