AI Coding
9 hours ago
- #AI
- #Programming
- #Compilers
- The author expresses skepticism about AI's ability to truly code, comparing it to a compiler that processes prompts (English) into code.
- English is criticized as an input language for AI coding due to its lack of precision, non-deterministic outputs, and non-local prompt effects.
- The effectiveness of AI in coding is attributed to poor existing tools (compilers, languages, libraries) rather than AI's inherent capabilities.
- AI is seen as a tool that could improve with better search, optimization, and pattern use, not as a replacement for human coding.
- The hype around AI coding is likened to past tech bubbles, with billions wasted on unproven technologies like self-driving cars.
- The author advocates for improving programming languages, compilers, and libraries instead of relying on AI hype.