Just a little problem with that: to understand the utterances of a vast number of humans you need to channel it to something that can understand the utterances of humans in the first place. Just channeling it around from statistic to statistic doesn't do the trick.