I dont understand the point the OP is trying to make. And the distinction between white and blue collars is unclear in this context.
Are mathematicians also blue collar since it seems they are being replaced as well? Are electricians white collars then since they are not?
Is knowledge work the hallmark of the white collar? It doesn't seem to stand to scrutiny: being an electrician seems to require a lot of knowledge, whereas some managerial roles do not in some cases.
All in all this seems like a pretty pointless discussion to be having. One of those language games people like to play (are we stochastic parrots as well?) and that dont bring anything of substance to the debate of ideas.
I am surpised that a lof of folks are surprised that neural networks work that well. I mean, did they really expect that we wouldnt approximate the brain cognitive capabilities at some point as we scale up compute??
This is rather obvious to me, but apparently not to everyone. My guess is that a lot of tech people have an immature model of the universe, have not read a lot of scientific, philosophical and cognitive literature. Or have not really taken the time to think or are not interested on such topics because they would rather spend time thinking about the nitty gritty and mundane workings of the framework du jour.
Neural nets are a toy approximation of biological neurons; and even if every synapse mapped to a weight (it's more complex than that but AFAICT even the brain scientists are not quite sure how much more complex), the biggest LLMs are around 0.2-1% of a human brain.
Shameless plug, AI built me this cli tool to visualize gh like diffs from the terminal https://github.com/trebaud/diffcat , could prove handy in times like these lol