ai-speed developer-productivity code-generation ai-tools

The gauge broke: devs felt 20% faster with AI, measured 19% slower

intrepidkarthi.com

July 2, 2026

6 min read

🔥🔥🔥🔥🔥

47/100

Summary

A controlled trial found that experienced developers felt about 20% faster while actually working 19% slower when using AI tools. AI increases typing speed but introduces overhead, prompting, waiting, and reviewing output, which can hinder productivity for experts in familiar codebases.

Key Takeaways

A controlled trial found that experienced developers felt about 20% faster using AI tools, but were actually 19% slower in measured performance.
The study indicated that the perception of speed among developers was misleading, particularly for experienced individuals working in familiar codebases.
Analysis of over 10,000 developers showed that while pull requests and code churn increased significantly, there was no net change in delivery, indicating a decline in delivery stability.
The increase in AI-generated code led to higher verification costs, resulting in reviewer burnout and a shift in bottlenecks from generation to review.

Read original article

View story →

Coverage & discussions

Hacker News · intrepidkarthi.com

The gauge broke: devs felt 20% faster with AI, measured 19% slower ↗

Community discussions

Hacker News discussion ↗

Historical count: 5 · 5 saved in sample · Historical sample

Analysis ready

Community Sentiment

Mixed

Positives

AI has significantly boosted productivity for certain repetitive tasks, allowing developers to focus on more complex and creative aspects of their work.
The integration of AI tools has the potential to streamline mundane coding tasks, which can lead to faster project completion in the right contexts.
AI's ability to gather context from various sources can enhance understanding and decision-making, making it a valuable tool for developers.

Concerns

The studies on AI's impact on productivity are flawed, as they fail to account for the variability in tasks and the changing nature of developer work.
Many developers report that while AI can speed up certain tasks, it can also slow down others that require deep contextual knowledge and understanding.
There is skepticism about the reliability of AI-generated code, with concerns that it may introduce errors and require extensive verification, negating perceived productivity gains.