Raw LLM Responses

​@Eunice3xyou say they can't "self reflect" but a little later on you describe them doing just that. Talking to yourself is pretty reflective and when you see what they say to themselves it's often quite reflective. I need to do this and respond like that to help the user. Also regarding the output not connected to the input, 2 things. There are models like that already. And what's the context if not connecting the output back to the input. If a person has the inability to form long term memories (it happens) and writes things down are they no longer a person? Their output is no longer connected to their input except via the written word. Pretty familiar sounding. Following step by step rules is a pretty arbitrary definition of consciousness. Are people who suffer discalculia also not sentient? They are unable to multiply numbers together either. These days when i ask it something like that it writes up a python script to get the answer right. The differences you describe are differences in quantity, not the presence or lack of fundamental nature. As AI improves i see people's definition of what is required to be sentient, sapient or intelligent becoming ever more narrowly defined to "must work exactly the way a human does" No I don't think AI "has a soul" or is truly self aware in a meaningful way just yet. But it seems like it is on the spectrum somewhere, it has abilities and intelligence that in some ways far exceed any given humans, and in others is far worse than most humans. I suspect many will never accept AI as being truly sapient so long as they know how it works and doesn't look like them. Pack the same brain up into a robot with a face and all of a sudden it's Mr data and it's ok.