I think the developer EGO is the most difficult part to manage during this AI development transition.
Like if "auditing" all lines produces by AI were meaningful, like telling AI hey I'm the boss I still have the last word.
I'm more inclined towards of building nice guardrails and deterministic checkers, like we did with compilers 40 years ago, are you still reviewing machine code to see if the compiler did a good job?
The same thing but a new level of abstraction.
commit 5980ac1