Unverified50% confidenceFactExact time
When an AI coding assistant directly produces code and submits a PR, the Git author field is often still recorded as the human developer who triggered it, causing traditional author-based statistics to be unable to distinguish human-written code from AI-written code.
1
Sources
50%
Confidence
Long-term
Relevance
10/15/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedTraditional Git workflows are built on the assumption that each commit corresponds to the explicit intent of a single human developer, an assumption that is broken by the participation of AI agents.74% similarUnverifiedGitHub的研究发现AI辅助编写的代码中安全缺陷引入率与手写代码相近,但因开发者对AI产出存在自动化偏见,这些缺陷更难在审查阶段被发现71% similarUnverifiedThe author divides AI responsibilities as follows: code is responsible for generating factual lines containing numbers, while Gemma is only responsible for generating advisory sentences containing no numbers whatsoever.71% similarUnverifiedAI代码贡献的核心问题在于贡献者可能并不真正理解自己提交的代码,从而将认知负担转嫁给维护者70% similarUnverifiedYou can use an AI input method, set the persona to "command-line expert," and describe your needs in natural language to obtain Git commands70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1038463API
curl https://kongchang.com/api/v1/knowledge/claims/1038463MCP
get_claim(id=1038463)