Linux Foundation research finds AI cuts coding time by half but increases 'AI slop' PRs

A new study by LF Research, titled "Navigating Open Source Software Development in the Generative AI Era," reveals a dual-edged impact of generative AI on the open source ecosystem. While AI tools can reduce code authoring time by up to 52%, they are also driving a surge in low-quality contributions, referred to as "AI slop."

Project maintainers are facing increased workloads as contribution volumes rise, with 59% reporting more activity since the advent of GenAI. The report highlights a critical gap in governance, noting that only 35% of projects currently have formal policies regarding the use of generative AI in their development workflows.

Productivity gains versus maintainer workload

• Efficiency Metrics: Generative AI reduces time spent on code authoring by 52% and proposal discussions by 19%, though 21% of respondents found it increased time required for high-complexity tasks.
• Contribution Quality: An estimated 25% of all incoming pull requests are considered low-quality "AI slop," a figure that climbs to 31% within large-scale projects.
• Governance Deficit: Only 35% of surveyed projects have formal GenAI usage policies, while 32% have no policy and no current plans to implement one.
• Contributor Behavior: While 61% of contributors perform functional testing on AI-generated code, the disclosure of AI assistance remains inconsistent across the ecosystem.
• Resource Reinvestment: Developers are utilizing saved time to address technical debt (66%), develop new features (63%), and engage in upskilling (58%).

Why it matters

The rise of AI-generated contributions necessitates new infrastructure for verification and automated testing to protect maintainers from burnout. Software engineers and project leads must establish clear disclosure and attribution standards to maintain community trust and ensure code quality.