I do find it amusing that “AI Bros” seems to have very little literacy of how the instruct LLM tools actually work, even at the executive level of companies. It reflects the sort of “confident ignorance” you see with being unable to look up documentation and such.
I find this frustrating because I consider myself a local LLM tinkerer and find it… Isolating. I love the problem solving loop and hammering gritty details out to coax decent tokens out quickly, or getting it to use documentation/forced syntax effectively, or assembling a bizzare dataset for some weird style or dialect, and that’s like antithetical to the culture.
I guess what I’m saying is there are problem solvers who use LLMs. Maybe they’re pink elephants.