生产环境中AI生成C++代码的质量特征分析

AI 摘要

一项针对大型企业生产环境中AI生成C++代码的大规模实证研究显示，AI生成的代码在接口耦合、拷贝分配开销和显式循环使用方面存在独特质量特征，导致审查工作量增加和5-8%的计算资源消耗上升。通过提供针对性的分类反馈，可减少11.1%的静态分析警告并提升计算效率。该研究基于2025年4月至2026年4月间352万次代码变更。

正文节选

Computer Science > Software Engineering Title:Characterizing the Quality Profile of AI-Generated C++ in Production View PDF HTML (experimental) Abstract:The widespread integration of AI coding assistants offers undeniable boosts to engineering velocity. Yet, recent studies point to a growing trade-off, revealing persistent challenges with code quality and maintainability. Industry leaders, including frontier AI labs, echo these concerns. As large language models are increasingly reli

发布时间：2026-08-10 12:00

抓取时间：2026-08-10 20:13

来源机构：arXiv