Andreas Kling Reflects on AI’s Changing Role in Open Source

Summary

In an X post, Andreas Kling presents the open-source community’s possible response to AI-generated code as a series of stages of grief. The denial stage dismisses AI-written code and assumes open source will continue operating as before. Anger focuses on low-quality AI pull requests and calls either to ban generated code or to reject the idea that projects remain open source when maintainers close such submissions. Bargaining accepts that AI can write code but seeks safeguards such as line-by-line human review or adversarial AI review. Kling then describes a depressive concern: if code generation becomes effectively free, the meaning of a contribution and the future of GitHub-style open-source culture become uncertain. His acceptance stage argues that open source can remain open while its contribution model changes, with issues, tests, benchmarks, specifications, and ideas becoming more valuable inputs. He also suggests that upstream projects may stop accepting arbitrary code while easier forking changes how participation works. In a bonus reconstruction stage, he proposes maintainer-controlled agents, issue-driven development, stronger testing and validation infrastructure, and new ways for contributors to earn trust. The post is a reflection rather than an empirical study or prediction backed by data.