That is exactly my experience. Every single time I get an LLM to write some code for me, it saves me no time because I have to review it carefully to make sure there are no mistakes. LLMs still, even after work has been done, completely make up methods and syntax that doesn't exist. They still get logic wrong or miss requirements.
Right now the only way to save time with LLMs is to trust the output and not review it. But if you do that, you're just going to produce crappy software.
- documentation for well-known frameworks and libs, "how do I do [x] in [z]?" questions
- port small code chunks from one language to another
- port configuration from one software to another (example: I got this Apache config, make me the equivalent in NGinX)
Which is already pretty cool if you don't think about the massive amount of energy spent for this, but definitely not the "10x" productivity boost I hear about.
Pretty much exactly this for me, except i can coax it into writing decent unit tests (really gotta be diligent though, it loves mocking out the things it's testing lol) and for CI stuff (mostly because I despise Actions YAML and rather let it do it). But i do get decent results in both areas on a regular basis.
Right now the only way to save time with LLMs is to trust the output and not review it. But if you do that, you're just going to produce crappy software.