Study: Human Code Review Limits AI Coding Gains
What happened
On October 10, Ars Technica reported that Harvard researchers Fiona Chen and James Stratton found AI coding agents generate more code but do not clearly increase software output. The study points to human code review as the efficiency bottleneck for AI coding tools, and says there is little evidence firms are raising software output or cutting headcount. The report separates code generation from software output: more code generated does not mean software output has clearly risen, and human review remains the limiting step the research identifies.
Written by AI from the coverage · updated 1 hour ago
Coverage
Follow the reports to see the story from different sides.
- Ars Technica · AIAI coding agents generate more code, but not more software
哈佛大学研究者 Fiona Chen 和 James Stratton 发现，人工代码审查成为 AI 编码工具的效率瓶颈，企业增加软件产出或减少用工的证据很少。
Heat over time
Not enough continuous observations to draw a trend yet.