AI research is like the topic of sex during adolescence: of everyone say they are doing it, few are; and the few that are in fact doing it are probably doing it wrong.