Let's make a matrix, x is Programmatic, basically how programmatic it is. A perfect example of a task that is not purely programmatic in software engineering is data normalization for multi-bank financial tracking. While you can write code to parse specific formats, the real challenge lies in handling the unpredictable, inconsistent, and highly fragmented way different financial institutions export data.

and put prevalence on y, how many people have encountered this kind of problem.

and maybe put risk on z, you wouldn't want a ai on a high risk job like surgery.

Let me think this through.