Raw LLM Responses

The big problem I have is that the official script is a summary of every answer this A.I made. So what he picks up is basically things which makes sense( The religion example he made with the Jedi order) from a big script of answer which makes most of the parts not really big sense. It's a question of propability. If there is a large data based feed for the A.I. and if there are thousands of thousands of answers there must be some answer which can indicate a sentient ability. The A.I. is not embedded in its environment than we are. He only is able to use fed data for a specific target we create. He can never use the data to represent it unlike we or animals do in our mind. Its like a pipe from beginning (data fed) until the end ( a target we created). The only thing which is new is that the A.I. is able to make new pipes.