Another difference is that you are predicting future sensory experiences in real-time, while LLMs "predict" text which a "helpful, honest, harmless" assistant would produce.