One Month Without AI
3 hours ago
- The author initially embraced AI coding tools, feeling powerful and productive, but soon became dependent and lost control over code quality.
- Using multiple AI agents simultaneously led to exhaustion, context-switching issues, and PR reviews that took days for tasks that could have been done in an hour.
- The author stopped writing code themselves, even delegating commits and PR descriptions to AI, and felt ashamed of not understanding the code being pushed.
- Quality of AI-generated code declined over time, requiring constant babysitting and prompting, reminiscent of lower-quality search results.
- A code review exposing a flawed test prompted the author to stop using AI, return to TDD, and reclaim understanding and confidence in their work.
- By quitting AI, the author regained joy in programming, efficient PRs, and the ability to justify every line of code, while maintaining their skills and value as a developer.