It's a tired discussion. LLM generated code can be awful, but it can also be very good. If you think all LLM generated code is bad, you haven't used any in the last year.
Company process and how LMS are treated in the development process is actually of interest. Are they YOLO shipping code or are they reviewing it and using LLMs as a tool, instead of a replacement for an engineer? That is where LLMs bring danger.
> If you think all LLM generated code is bad, you haven't used any in the last year.
I use Opus 5.5 and it can't generate code anywhere near as good as mine. I'm not an exceptional programmer either, I'm fairly middle of the road. LLMs remain bad at programming, despite the claims of AI bros that only within the last 6 months have models become good (which they curiously have claimed constantly for years).