@baldur
A thought experiment I like to mull over: if LLMs had been available before high-level languages, trained on •assembly•, and if we take as a given the extremely dubious assumption that people go all in on LLM-based coding for all those years, what would programming look like today?
The answer I keep coming back to is that code would be utterly impenetrable to humans, almost as opaque to us as the human brain but far less reliable.