In a code-driven software paradigm, programming languages are the principle means to instruct machines. This includes AI approaches that use written code as the source for modeling how to generate code. Code created this way imposes costs and risks associated with developing, testing, integrating, deploying, maintaining, and securing it.