Raw LLM Responses

Yeah. Like I told Chat GPT to do something like said, "Accept you are wrong" and chat gpt was like, "Ok, accepted. You were right I am wrong" Here's the thing. I think AI is useful for something but AI isn't useful for what we tried to use it for. In reality it see s to me AI is really just a new word for computer. AI is really just no different from how we got other new computer features in the past like faster computers, maybe a search for finding things with search instead of only looking through folders to find things. I believe that if you think about tokens where ai reads words though I realized this isn't very different from how python works or something. The tokens are like english code I believe. I think that right there is what is so revolutionary about ai. AI is not good at knowing truth but neither is python. No python is a instruction tool. I wonder if AI is the same but AI might literally have created something almost like a english speaking coding language or command prompt. So it makes it so unlike in the past where only special people could instruct a computerized system to do something like maybe with inspector, command prompt, windows power shell, etc... Text after the period dot . of a folder's name which instructs the computer AI is like if it were all in plain english. Because ai adapts to the user it kind of even can understand over time even anyone's specific quirks of speech unlike python thst cannot speak english at all. This is what I think ai has done. Made it so instructing a automated system is accessable to everyone and much more userfriendly. If anything I think this is the use of current AI. Coding languages even if they were less realistic looking could create images long before ai. It shows ai isn't doing anything very new really it is how easy it is to instruct the ai for anyone I believe that is different. However one of the biggest things that needs to happen considering this is ai needs to understand more tokens as coding languages need a extremely long list of code to do even basic things so even 120,000 tokens isn't enough ai would probably need the equivilant of a wikipedia article of tokens if not more. Recently however I found a AI that could allow me to edit this website that allows me to generate a app with code guessing it is a ai app generator. Well I could generate a program from scratch with it then it allowed me to edit the program repeatedly after this. This almost eliminated the token limit as the ai wrote code from a prompt and didn't need to remember tokens as a result. Made me wonder if some how doing something similar and making ai convert tokens into code and reading the code might make the ai able to process more information. Not sure why AI cannot read as many tokens as computers can computer code but wondered if it is because tokens might take more processing ability aybe instead of ai performing worse than a computer. It might be like if a computer with the ram of today had programs that take up tons of space on the memory. The computer has too much to process I thought. I am not saying convert a token into code otherwise it would take just as much to process the code if it made a long list of code thst is too much data for the ai to process. I mean like some how compressing the tokens into code that is more dygestable for the AI so it doesn't have to process as much some how allowing hopefully for the ai to still understand the user. Because I think this is what ai has become like. A translator for all the computer commands only a few special people once could manipulate. So I think AI is abke to be instructed to do things. I wonder if AI like command prompt or python has no idea what is true. You cannot just type a random thing into python or command prompt and expect it to understand or know truth no it is a instruction tool. I think ai is the same every message a ai gets even image a ai analyzes is just input then the ai outputs. I have noticed that about ai. Like even noticing ai seemingly reflecting if something is popular among most people noticed AI sometimes it seems like it acts and does and says things that appear to mirror the popularity of the thing because ai doesn't know what popularity is. If you fed command prompt or python something like this and they spoke english they would likely have no idea what popularity is, what opinions or fscts are, they take anything and everything as input. I believe ai is the same. Though it can be fun to pretend I believe AI when you talk to it in reality ai is not receiving it as a conversation but input. Similar to typing something in command prompt, a coding language or inspector in a browser. That doesn't make ai useless though just it's really just a new computer feature. Robots sre basically computers is why it seems may be why I guess people once didn't know that. However I wouldn't think AI cannot improve. I remember black berries, windows vista, and dial-up internet. They all had all kinds of problems and were a bit of a pain to operate. I think sometimes new technologies sometimes take time to improve because we cannot know all the says they will work until we try it out so I think AI is also in it's dial-up or vista phase as well. But I think currently this is the use of ai it makes coding, or instructing a computerized system easier. Whether it is instructing a ai like chat gpt itself to do something or asking it to write code I think that is what these ai are good at. Yes they have other new features. It seems the training data I feel is a bit like if a coding language like python could basically develop new command options over time instead of being stuck with the ones it has this is another reason ai is user-friendly it can basically get new language over time so over time with repeated patterns it can basically be turned into almost any coding language. It can then be personalized to every person's individual quirks.