Аннотация:The authors review the structure of definitions and foundational standards being developed by the International Standards Organization for systems with artificial intelligence. The authors demonstrate that these definitions produce results which depend on the set of states of the world where the system is used, ie. the results are domain dependent. The authors further propose that such domain dependence is a universal feature of weak AI systems and a measure of domain dependence may be used to haracterize the strength of a particular AI solution.