> So my first hand gut reaction is that people who cannot fathom meaningful use of modern LLMs are by definition people who are not trying to solve complex problems in domains they aren't yet super confident in.
OTOH I would never trust anything built by someone not super confident while heavily relying on a LLM.