News Summary

Recent studies have cast doubt on the effectiveness of AI coding tools in enhancing developer productivity. According to a research conducted by METR, a non-profit AI research group, developers using AI tools like Cursor and GitHub Copilot experienced a 19% increase in task completion time instead of the anticipated 24% decrease. The study involved 16 experienced open-source developers tasked with completing real tasks on large code repositories. Despite prior experience with web-based LLMs, only 56% of the developers had used Cursor, the main AI tool in the study. Potential reasons for the slowdown include the extra time spent prompting AI and its struggles with large, complex code bases. While the study suggests skepticism towards universal productivity gains promised by AI coding tools, it also acknowledges significant AI progress in recent years, indicating that results may differ in the future.