Torvalds Says He Likes AI as a Tool but Warns Kernel Use Needs Care

Torvalds drew a line between hobby use and critical development. “I think that you need to be very careful using AI when you’re doing something real and important,” he said, according to ZDNet. He described himself as a maintainer rather than a heavy kernel programmer, saying the Linux kernel is a project where other people do the real work and he acts as a collection point.

Even so, he said he still enjoys programming immensely and considers AI a useful tool if used correctly. “I think AI is a wonderful tool if you use it correctly and treat it as a tool, and I find it makes programming much more enjoyable,” Torvalds said. He recalled starting programming in 1981 or so, when computers and programs were simpler. Today, he said, the bar in software engineering has grown so high that new developers can struggle to see their small efforts as worthwhile compared with polished professional programs.

Torvalds said AI can help beginners find joy in coding. He defended vibe coding as a way to make programming feel relevant and to let new programmers do things that would otherwise be difficult. He described AI as a “gateway drug,” citing his own toy project, a guitar pedal. He had written a user interface in C for a microcontroller and a small screen but wanted it to look more modern; because he does not do Java, he turned to AI for tasks he is bad at. He said his first AI attempt was “horrendous,” then he showed the AI his existing C approach and told it the result was already correct but ugly.

In the Linux kernel, AI is already part of the workflow. Hohndel asked about Sashiko, an agentic Linux kernel code review system. Torvalds said its public reviews now appear on the Linux Kernel Mailing List, and some subsystem maintainers expect patches to have received such a review before they are accepted. Not all pull requests must go through an AI review, but some maintainers require it, and Torvalds suggested it is likely all will eventually require it.

Torvalds said AI tools can identify genuine security problems and surface issues in old, no-longer-used drivers that had gone unnoticed for years. He sees value in the resulting fixes but acknowledged the cost. “It is, I think, improving our code base… Although it is also stressing maintainers out to the point that it’s a problem,” he said in the ZDNet-reported conversation.

The strain is not theoretical. Torvalds said in Mumbai earlier this year that convincing but fabricated bug reports can require substantial human effort to disprove, while narrowly targeted patches may fix one symptom without addressing the underlying problem. He called some submissions “mindless band-aid kind of patches.” He also said three-quarters of the preceding day’s Linux Kernel Maintainer Summit discussion concerned 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.

Torvalds’s comments came amid continued resistance to AI in open source. ZDNet noted that System76 recently banned AI-generated content from its COSMIC desktop. Still, Torvalds said AI code checking is here to stay.

Because it was Linux’s 35th anniversary, Torvalds also discussed how the process of making Linux has changed. In the beginning, he said, Linux had no development infrastructure to speak of. He tracked changes through tarball patches and regularly distributed updated source code, making patches every day internally and releasing updated source code regularly.