@Hitspark So I hate to be that guy, but no-one "drove a robot to want to commit suicide". The sort of program used was a very sophisticated kind of linguistic mirror, designed to spit out answers to questions asked as a gramatically-and-context-matched version of however people have answered...