A new study just debunked the biggest fear about AI and open source (thenewstack.io)

A recent study from Peking University challenges the prevalent fear that AI coding agents hinder newcomer participation in open-source projects. Analyzing 1,888 GitHub repositories that adopted AI tools like Cursor and Claude Code, researchers found that newcomer involvement remained stable or slightly increased after AI implementation. While there was a modest rise in complexity—3-4% in cyclomatic complexity and 11% in cognitive complexity for Python projects—the study suggests that this increase did not deter new contributors. Conversely, the previous Carnegie Mellon study indicated a more significant rise in complexity due to AI use, highlighting the need for more comprehensive analysis in varied project contexts. Significantly, the study focused on established projects, leaving out new ones that adopted AI from inception, which could skew results. Additionally, it measured adoption based on configuration files rather than actual usage, limiting insights into the impact of heavy reliance on AI tools. Despite these limitations, the findings point to a nuanced picture: while AI-generated code is more complex, it isn't discouraging newcomers. However, the influx of more complicated contributions increases the burden on maintainers, whose numbers are not growing at the same rate as pull requests. This raises critical questions about the future demands of maintaining open-source projects in an AI-enhanced landscape.