A new paper suggests that vibe coding, the practice of using LLM-backed chatbots to generate code, may be detrimental to the open-source ecosystem. The authors warn about reduced user interaction with OSS projects and the difficulty of starting new ones. The reliance on LLMs could affect the selection process for libraries and tools.

The shift towards AI-assisted development could turn developers into customers of chatbots, removing the need for them to understand the generated code. This process can also eliminate the usual selection process of libraries and tooling, replacing it with what's most prevalent in the LLM's training data. This could lead to a decline in community forum usage.

The paper highlights the potential problems if developers delegate actual engineering to LLMs. LLMs won't interact with library developers, submit bug reports, or be aware of potential issues. While the impact remains uncertain, the authors suggest software ecosystems like JavaScript and Python could suffer most.

This trend follows the launch of GitHub Copilot in 2021, with reports indicating limited benefits and increased bugs. As AI-powered code generation evolves, it's crucial to consider the impact on developers' cognitive skills and the long-term health of open-source projects. Many are concerned about the future of AI in software development.