Then explain to me how the neural networks behind LLMs are not mimicking neurons. Because that’s the basic idea behind all of deep learning.