Dev Tools · 2h ago
Study: AI Coding Assistants Can Slow Developers by 19%
A METR study found experienced developers were 19% slower when using AI tools on their own repositories. The productivity loss comes from time spent prompting, reviewing, and refactoring AI-generated code. The article argues that benchmarks don't measure real-world productivity and that developer skill matters more than model size.
Meridian48 take
The 19% slowdown figure challenges the prevailing narrative that bigger AI models automatically boost developer velocity, but the study's narrow scope—experienced devs on their own repos—may not generalize to all coding tasks.
ai-productivitydeveloper-tools