Raw LLM Responses

I'd like to strongly state - There is an unspoken confusion about what we believe this current AI is, VS what we have been shown as Autonomous Intelegence (sci-fi). One thing that spoke loudly was when the Google staff said "You need to be able to prompt the AI effectively, to make sure you get the desired output you want."  To me, this sounds like “a glorified function”, still takes the “input” and yes can or cannot have fixed interpretations in the middle to generate a final response with, then output that response, VS artificial intelligence what is presumed by science fiction is the ability to act on one’s own accord to facilitate the desired notion, action or request by any means possible. Whereas what we seem to have in our current reality is a system that is fully contained, siloed and fed information rather than it building information on its own. Also each “AI” has a context window, each window can only go back so far in its memory to tell you what the context is, and sometimes while communicating, the context can be altered by the AI due to the limited context window or memory.  This is why I say we seem to be confused, and quite possibly jumping the gun saying this is all "AI", when in-fact is nothing more than a complex function/prompt interpreter/context generator.