Linux memory management reviewer steps down amid AI slop patches

Lorenzo Stoakes, a lead Linux reviewer for memory management, has announced that he will step down from several review duties in that area, while keeping other responsibilities such as KVM. He said the decision followed mounting pressure on his inbox, including a growing share of AI-generated patches he described as slop. The announcement was made today in a thread on the Linux kernel mailing list.

Which duties are changing

Stoakes will leave the core, reclaim, Rust, THP (Huge Pages), and miscellaneous sections of kernel memory management. He will stay on for his remaining duties, including KVM. He plans to wrap up the patch series he is currently working on and occasionally pitch in. The partial departure appears amicable, and the developer does not appear angry or bitter.

AI slop and review workload

Stoakes said the AI wave has been horrible and is getting significantly worse. He described a progression from occasional slop to at least once a week, then every few days, and now pretty much half of what he sees, if not more. He had raised concerns back in July about a recent wave of slop in memory management and said the amount of time wasted on AI slop was already too much and only at the start of this. As recently as a month ago, he said what kills the project is having 1,000 slop submissions that 2 or 3 people have to do something with.

Open source and AI contributions

AI-assisted programming has arguably been both a boon and a curse for open-source software. Open projects let anyone find or report bugs, fix them, or add functionality. That openness also means anyone can contribute, which is suboptimal in a world with low-effort AI slop and AI agents. Open-source projects are hardest hit because they rarely have adequate manpower or funds to keep code up to date, much less deal with large volumes of machine-generated submissions.

Linux stance on AI-assisted code

Linus Torvalds, creator and headmaster of Linux, has taken a pro-AI stance. That stance is supported by the speed of additional contributions and by the fact that bots often do not make mistakes that even seasoned developers do. The topic has been the subject of a lot of LKML discussion, and the current stance is that AI-assisted code commits do not need to be marked as such. The problem has become one of filtering, since accepting AI contributions is easier when reviewers are not wading through mud to find one gold nugget.

Sources: Tom's Hardware