#llm #ai Created at 110423 # [Anonymous feedback](https://www.admonymous.co/louis030195) # [[Epistemic status]] #shower-thought Last modified date: 110423 Commit: 0 # Related - [[Computing/Analogy LLM context and human thinking]] - [[Computing/Intelligence/LLMs are a lever rather than an engine]] - [[Computing/Automatic RLHF]] - [[Computing/The advantage of LLMs to understand human memetics]] - [[Computing/Intelligence/Evaluating LLM]] # Context locality in LLM What if [[Large language model|LLM]] could take infinite size input? Is it even possible ? How can it even predict or solve a problem given infinite input? You at least need a goal right? Aren’t human being especially throwing useless messages information away, [[Philosophy/Rationality/Models/Noise|noise]], to keep only [[Signal]]s, what is aligned with your [[Personal growth/Goal|goal]]s? [[Large language model|LLM]] will always need to be fed the right window of context and that’s why [[Embedding is the dark matter of intelligence]] ## the non local nature of the [[Quantum physic]] [[Nicolas Gisin - Quantum Chance - Nonlocality - Teleportation and Other Quantum Marvels|Quantum Chance - Nonlocality - Teleportation and Other Quantum Marvels]]