AI Lazyslop and Personal Responsibility
16 days ago
- #code review
- #personal responsibility
- #AI in coding
- A coworker (Mike) submitted a 1600-line AI-generated PR without tests and expected immediate approval.
- Pushback on adding tests led to frustration and sneaky merges of changes into approved PRs.
- The author emphasizes personal responsibility when using AI, urging thorough review and disclosure.
- Key questions for AI-generated code: thought process, prompts, personal review, and confidence in functionality.
- Suggests adding comments to PRs to explain the AI-assisted process and using AI reviewing tools.
- Highlights the importance of writing meaningful tests, not just superficial ones.
- Discusses the cultural shift towards AI in coding, with examples from Ghostty and Linus Torvalds.
- Defines 'AI Lazyslop' as AI-generated code not reviewed by the creator, burdening reviewers.
- Proposes an 'Anti-AI Lazyslop Manifesto' advocating for ownership, disclosure, and thorough review of AI-generated code.
- Mike evolved to 'semi-lazy-slop' mode, using AI to address reviewer comments, raising questions about its effectiveness.
- The author discloses using AI (Claude) for grammar and style checks in the blog post.