Software craftsmanship emphasizes well-crafted code, individual responsibility, and programming mastery. In AI-assisted development, its central challenge is preserving human judgment when models produce code faster than teams can verify it. section 1.8
Disciplined practice matters because generated code is unevenly difficult to trust:
- Use models for scaffolding, boilerplate, and common patterns, while treating edge cases, integration, security, performance, and verification as human responsibilities. section 1.8
- Improve specification quality rather than assuming a different model will remove the need for judgment; model-generated review can share the original model's blind spots. section 1.8
- Treat ownership as understanding and validating the whole system, not merely producing individual parts. section 1.8
