What's the relationship among LLM, Prompt, RAG, Prompt Engineering, Metadata?

the actual reason for prompts is that LLMs are trained on massive amounts of data. so its not just that you are telling it to look for specific data. its because the LLM dosent actually read words. it reads the semantic values of words. every prompt you send is broken down in to tokens, then processed in the semantic cloud, so you can envision your text as a ‘cloud’ of sorts. the LLM is now trying to match the ‘shape’, ‘texture’, and particular ‘hooks’ of your ‘cloud’ with other clouds. the fact that this works at all is amazing. the fact that it works over and over again across models is a mundane miricle. but thats why prompts.