When designing in A.I. you straight find that any apparently trivial problem always takes a complex math-conception and programming. Heuristics, Recursivity and massive data are at the very core of it. So, it means that a carefully optimization of algorithms is essential, in terms of structure and equivalent formulas with very low relative error (ε).
So, strategically speaking we’d say that is an optimum point in which the net value of the Artificial Intelligence Program (AIP further) is maximum. In a graphic:
