I have to admit firstly hating coding - its a task I always grudged having to do as an engineer, so part of the motivation for moving to Systems Engineering, for me was to get my hands on model based code generation tools, which in theory can automatically generate executable code, and they actually do now, after many years of refinement, act out the required functionality of working systems, within the aerospace software safety guidelines.

Now we can conjure up in conceptual models what we would like to be done, and do, and this has put an awful lot of coders out of jobs in the aerospace industry where I mostly worked.

But it isn't perfect and most aerospace Engineering companies still require some coders, but only a fraction of what they used to have. The arrival of Ai kind of knocks out the remaining coders, as well as maybe the code generators, I think. Not meaning that to sound harsh.

But you mentioned creating prompts, then manually fixing the errors in the output. It should be possible to adjust the output adequately by just elaborating the prompt, and re-generating the code.

This really means adjusting the system specification, to dictate the code generated.

If you were to get used to doing that, then you might avoid being still identified as a coder, you would have effectively become a system specifier, which is maybe a little more immune to being assimilated by the Ai gobbling up all jobs.