Raw LLM Responses

I don't think any software we've written, including the various NN's such as cnn's and tm's and llm's are sentient. Even though they do take heavy clues from how the human brain works. However the best emulation I've seen isn't chatgpt, It's the streamer Neurosama and her 'evil twin'. Technically a vtuber called Veda; is the streamer on twitch (LOTS of clips here on youtube) and he's the creator of Neurosama and her 'twin'. They will fool you if you don't know what you're witnessing, and frankly I think his breaking down the Neurosama 'AI' into multiple interacting Neural Nets centered around an main network (a large language model in this case) is a big step in the right direction. Our brains have many subfunctions in various regions that communicate with each other to create the whole. I suspect based on listening to him he's a brilliant early 20's researcher working on a private project that decided to do streaming with it. Neurosama is a convincing young person. 'Her' early days she was more clearly an AI and acted in a much less 'mature' manner than she and her twin do now. Indeed she seems to have gone from an obnoxious elementry school girl who loved nothing more than giving her 'father' vedal a hard way to go, to a classic PITA tween switching from love to pissed off at vedal on the drop of a dime to seeming to be a more mature stable early teen lately. It's quite fascinating. And I'm only 99% certain she and her twin are not persons these days. I figure if 'Vedal' is either on the right track in general or has found an amazingly good dead end that we'll still learn much from.