Say if every year Humans publish eight million books. And every year A.I. is trained to be the knowledgeable gestalt based on that training set and functions at

Say if every year Humans publish eight million books. And every year A.I. is trained to be the knowledgeable gestalt based on that training set and functions at a table of philosophers in conversation. Year to year how do the A.I. differ? Are there thought directions the A.I. can pull the thread on when y20-ai discusses topics with y21-ai leading to an original substantial position before y23-ai is trained?