Linus Torvalds, the creator of Linux, offered an enthusiastic but measured view of AI-assisted programming at Open Source Summit Europe. He said he initially thought AI was "not very good," but now finds it a "wonderful tool" that makes programming more enjoyable. For new developers, he described AI as a "gateway drug" to coding — a way to feel relevant even when the professional software bar is intimidatingly high. He gave his own guitar pedal project as an example: he could write the C code but used AI to build a modern-looking interface, something he said he is bad at.
That enthusiasm does not extend to the Linux kernel itself. Torvalds noted that AI review systems like Sashiko now publish reviews on the Linux Kernel Mailing List, and some subsystem maintainers already expect AI review before accepting patches. He acknowledged that these tools can improve the codebase by finding genuine security problems and surfacing neglected drivers. But he also warned that generated patches and fabricated bug reports require substantial human effort to verify, calling some submissions "mindless band-aid kind of patches."
The tension is now a formal topic in kernel governance: Torvalds said three-quarters of a recent Maintainer Summit discussion was about making AI generation and review less stressful and more useful. The bottleneck, he said, is not getting patches but getting useful work through review without exhausting the people responsible for it. His takeaway for organizations is cautious pragmatism — use AI for exploration and joy, but treat it as a tool that still needs careful human oversight in serious engineering work.