It not really a turing test, its like probing the "associative memory" in a huge database for compatible strings, except that memory is just a map of indexes to

It not really a turing test, its like probing the "associative memory" in a huge database for compatible strings, except that memory is just a map of indexes to indexes. You need to use more abstract and non-sense stuff to uncover the raw mechanism, but its still statistics not "understanding" what these words are(they're just indexes with high correlation to other indexes) and stringing them into run-on "generic templates".