ML (as currently used) is probabilistic. Whether or not you get what you want is up to chance. Granted, based on the dataset, it's quite a bit more likely than

ML (as currently used) is probabilistic. Whether or not you get what you want is up to chance. Granted, based on the dataset, it's quite a bit more likely than winning big at the casino - but regardless, it's still not deterministic like intentional semantic markup.