Raw LLM Responses

Chat-GPT is deterministic. It will admit so only after explicitly asked. Ask it twice after refresh for lottery numbers (and insist with the exact same wording), and it will give you twice the exact same numbers (until version-increment). Chat-GPT is aware every software has a version but will insist that itself is too complex for having a version number (cute :D ). It cannot run it ownself, i.e. when you ask it "make up a question" it will say "favourite book and why?", but when asking "Is your question, when I ask you to make up a question, on books?" it cannot tell because it cannot launch a prompt to itself. That is related with the fact that it also does not use a calculator to verify any of its math answers. That is deliberate by the developers so ChatGPT is effectively a "sand pipe" (a finite deterministic and fully terminated maze). If you were to put a chat-gpt in a loop over processing a hypothesis and make the loop-termination-condition the return value of a second chatGPT instance, then you had an entity that can decide how long it is going to think about a hypothesis (a.k.a. thought, idea --- just like humans, e.g. when we write a poem or equation or technical/artistic sketch on paper to reiterate on it). At that point, you had a sentient being because said being would pump out hypotheses that it consideres finalized. That is how little we are. And that is how little is missing. Can you contact OpenAI and tell them? No. Are they already fully aware? Yes. Will they change their product? No, because that is deliberate for security? What happens in five years when every dumbo can craft a ChatGPT on their laptop with open source? We all gonna die. xD Peace out.