NewsBite analysis · AI, Programming
MacRumors
AI coding tools make developers slower, study finds
Researchers from Model Evaluation & Threat Research conducted a study involving 16 experienced developers to gauge the impact of AI coding tools. Contrary to their expectations, findings indicate that these tools lead to a 19% increase in task completion times due to factors like high familiarity with workflows and AI unreliability. While developers expected a speedup, the results suggest additional time spent on managing and validating AI outputs renders the use of these tools counterproductive overall.
MonitorConfidence: HighMid term horizon
Published 27 Jul 2026, 14:45 UTC4 min read
Visit original article ↗Lead image — MacRumors
Summary
TL;DR
- AI tools perceived to increase efficiency actually slow down developers.
- Study observed a 19% increase in task completion time with AI assistance.
- Developers relied on AI-generated suggestions less than 44% of the time.
- Familiarity with code repositories contributed to tools being less helpful.
- Findings do not negate the potential future improvement of AI tools.
Interesting facts
- Despite expectations, the study revealed AI tools hindered productivity.
- The study involved 246 issues from real open-source projects.
- Existing literature supports multiple claims of AI tools not delivering expected benefits.