Linus Torvalds is using AI to develop Linux kernel code now (www.zdnet.com)

Linus Torvalds has taken a significant step in the development of the Linux kernel by using AI, specifically Anthropic's Claude Code, to write a new set of patches aimed at addressing a long-standing memory cache slowdown issue. This innovative approach marks a shift in Torvalds' perspective on AI, transitioning from skepticism to embracing it as a valuable resource for programmers. The newly proposed patches tackle contention in Linux's page cache, which can severely impact performance when multiple threads access the same cached file. Through new timing strategies, the changes aim to optimize how memory is reused, leading to impressive performance improvements in specific workloads. The implications of this announcement are profound for the AI/ML community, as it showcases a major figure in open-source development leveraging AI to enhance complex system performance. While initial benchmarks on Torvalds' laptop indicate substantial gains—such as increasing throughput from approximately 10 million to over 30 million reads per second—challenges remain regarding memory management and possible regressions. Torvalds emphasizes that while using AI is exciting, caution is necessary, especially in critical development areas like the Linux kernel. This moment represents a pivotal shift in how AI can assist in serious programming tasks, potentially influencing future development practices across the tech industry.