Anon11/29/25, 21:10No.16859720
LLM like gemini 3 and chatgpt5.1 and grok are just very a bunch of multidim manifolds of vector amplitudes, when we use them it's just input being "tokenize" and goes into this manifold to retrieve informationthese are pulses, of a chained up meaning, each time you interact with it, there is no feedback no awareness, no qualia, there cannot be consciouness because it is only enough pulse to satisfy your answer, calculated in tokens
they are more like snapshots, between the vector space and your tokenized inputso the gpus or server clusters are basically manifold generators, the training weights (this bector space) is dormant, image a huge text file of matrices, in order to "travel" through it the gpu act as a projector projecting the surfaces for the pulse to travel throughllm behaves like a jellyfish, no brain just pulses, but it travels in a temporary multidimensional tunnel