Comment by skort
8 hours ago
I feel like if anything LLMs have reduced coding/software to "throw as much as possible at the wall and see what sticks". It feels quite shortsighted and wasteful, especially considering that humanity needs to get better about how it produces and consumes energy. It's sort of bleak.
Which is a bit terrifying and removes the ‘engineering’ component of a critical part of the job, in exchange for slot machine / pray to ai gods…which comes with disconnect from the code, something I assume will adversely compound over time.
For work I'm still responsible for what the code does, so while we use them to write code, we still review, understand, and change how it does things. Because oftentimes even the latest models do dumb ass shit.
But everything I've vibe coded with it is a massive pile of shit. I look under the covers and its just hack built on hack that solves the most recent requirements problem but falls over in a stiff breeze. If you thought "this should be easy" was bad for human coded software, vibe coded software is a whole other level of holy shit.
The most recent WTF was a changelog feature. The LLM had repeated the max id used in 3 different places. It missed a place during an update and later it confused itself.
hasn't improved software either. it's made the overall quality worse if anything.