I see your positive view on programming and raise you this: https://www.stilldrinking.org/programming-sucks https://www.stilldrinking.org/programming-sucks The

I see your positive view on programming and raise you this: https://www.stilldrinking.org/programming-sucks https://www.stilldrinking.org/programming-sucks The fact of the matter is, that a lot of the development work out there is just boilerplate: build scripts, bootstrapping and configuration, defining mappings for Web APIs and ORMs (or any type of DB interaction), as well as dealing with endless build chain errors and stuff I honestly think is bullshit. When I see a TypeScript error that's borderline incomprehensible, sometimes I just want to turn to an LLM (or any tool, if there were enough of formalized methods and automatic fixes/refactoring to make LLMs irrelevant, I'd be glad!) and tell it "Here's the intent, make it work." It's fun to me to dig into the code when I want to reason about the problem space and the domain, but NOT very much so when I have to do menial plumbing. Or work with underdocumented code by people long gone. Or work on crappy workarounds and bandaids on top of bandaids, that were pushed out the door due to looming deadlines, sometimes by myself 2 months prior. Or work with a bad pattern in the codebase, knowing that refactoring it might take changes in 30 times that I don't have enough time for right now. LLM makes some of those issues dissolve, or at least have so little friction that they become solvable. Assumption: when I use LLMs, I treat it as any other code, e.g. it must compile, it must be readable, make sense and also work.