Thorsten Ball argues AI shifts software's bottleneck from coding to problem definition

Thorsten Ball argues that when AI model code quality exceeds human review capability, established processes centered on human coding will collapse. Code review, unit tests, and 'good code' standards serve human readability and lose relevance when code is primarily modified by AI. The bottleneck shifts from writing code to defining the problem correctly, making the builder's craft - understanding business problems and feedback loops - more important than coding craftsmanship.