'Human In The Loop' is not enough
3 points by stilic
3 points by stilic
This feels over reactionary tbh. We've been automating aspects of code creation since we began writing code. It's almost like writing code was never the actual job in the first place.
I have become increasingly cautious of claims like "writing code was never the actual job in the first place", because if writing code was not, then what else was? Does this come from an unbiased position or is it a coping argument for using AI? Some would claim it is even insulting.
Secondly, there is a distinct difference between automating aspects of code, and automating code itself. It is similar to automation in the kitchen, using a food-processor or a mixer-grinder doesn't mean the entire meal is machine-made.
"Human in the Loop" also inherently causes burnout: GenAI models can output so much code, you can't expect someone to read it all and actually understand it, unless it is very rote. Everyone saying that they "read all the code" is either just starting out and still has energy to do so, or is not actually doing that.