There's been a lot of criticism thrown out lately about the gradual ceding of ground to AI programming agents within open source projects -- notably within crucial digital infrastructure such as the Linux kernel, and various system utilities or tools.
I'm curious. Has anybody with the expertise to do so gone through various commits to open source projects, in order to analyze the good/bad within these codebases since introducing AI agents to the process?
I feel like a resource that actually dissects *real code* -- not just theoretical concerns about tech debt -- would be useful in the overall discussion around this stuff.
I know it's somewhat dubious to even properly try to figure out with 100% certainty which commits are produced by AI agents vs which were handwritten unless the agent specifically is included in the git commit. But, I still do wonder if anybody has done anything like this.