Still dont understand the push back against models writing code
retained text
Genuine question: What's the real problem with AI writing code? I'm trying to understand the criticism surrounding AI generated code, particularly the argument that it produces nothing but garbage code. Based on the tasks and projects I've been working on, AI has helped reduce what could have been months of manual work into days or even hours. The productivity gains are difficult to ignore, especially for someone who is still learning and doesn't have years of development experience. I understand that AI can produce inefficient, insecure, poorly structured, or unnecessarily complicated code. I'm not arguing that everything it generates is ready for production. But isn't that where reviewing, testing, debugging, and refactoring come in? Are people using AI generated code without understanding or reviewing it, or is there a more fundamental problem with the code itself that I'm missing? To me, AI is a tool that accelerates implementation, but engineering judgment is still necessary to validate the output and make sure it's maintainable, secure, and fit for purpose. So, for those of you working professionally in software development, what's your actual experience? Is the criticism mainly about how people use AI, the quality of the code it generates, or the long term maintenance costs that come with it?
document fetched 2026-10-26 00:36 · canonical https://reddit.com/r/ClaudeCode/comments/1xfag1h/still_dont_understand_the_push_back_against