> I feel with LLMs, we're still very much at the "measure the elextrical charges" stage I think that's a great example. Say you somehow had a skilled electrica

> I feel with LLMs, we're still very much at the "measure the elextrical charges" stage I think that's a great example. Say you somehow had a skilled electrical engineer analyzing how a little IC manages to spit out a sorted array when given an unsorted array, but you're in a world without computer science or even information theory and this IC had just come through a portal. How could they figure out the explanations that we have about programming languages, compilers, assembly code, transistors, etc.? Well, they'd probably have to invent information theory and computer science before they'd even know what such an explanation could even feasibly look like.