A recent study conducted by the nonprofit research organization METR has found that the use of AI coding tools can reduce the productivity of experienced software developers. The study tracked 16 veteran developers as they addressed 246 real GitHub issues within large codebases exceeding one million lines of code. Results showed that when AI tools such as Cursor and Claude were allowed, developers worked 19% slower compared to when they did not use AI assistance. The research highlights that AI may not accelerate development for complex, familiar coding tasks and could even hinder efficiency in certain contexts. This finding challenges the prevailing assumption that AI coding tools universally boost developer productivity. The study has attracted attention across multiple technology news outlets, emphasizing that while AI can be beneficial in some coding scenarios, it is not a guaranteed solution for improving software engineering speed.