Ask HN questions AI coding productivity declines

news.ycombinator.com2 min read

The author questions the actual productivity gains of AI-assisted coding, arguing that as codebases grow larger, AI models experience drift, inconsistency, and code collapse. Despite various mitigation strategies like plan modes and automated testing suites, the author notes that tests often become self-confirming while underlying bugs persist, mirroring broader studies showing little to no productivity gains for experienced engineers. Open article →