- Linus Torvalds says he doesn’t recommend AI-assisted coding for kernel development
- V”ibe coding” helps beginners perform tasks they cannot complete independently
- Torvalds says vibration coding creates long-term maintenance issues in core systems
Linus Torvalds discussed the use of AI-assisted coding, proposing a measured stance that separates experimentation from development from production.
The godfather of Linux acknowledged that some developers may already be exploring such tools for kernel work, although he is not personally using or testing them.
Torvalds claims that AI tools have caused disruption through crawlers harvesting kernel.org source code, leading to fabricated reports of vulnerabilities and bugs.
“Vibe coding” as an entry point and not as a basis for critical code
Although Torvalds noted that these issues affect the Linux kernel, he suggested that the situation is more serious for other projects, citing curl as an example.
Torvalds described himself as favoring “dynamic coding” when it helps users learn programming or perform tasks they couldn’t accomplish without help.
He compared this with his early experiences typing programs from print magazines, saying that computing is now much more complex.
However, he reiterated that this approach should not be applied to core systems like the Linux kernel, where it could create long-term maintenance issues, and suggested that using vibe coding in such environments would be a bad choice, even if it seems attractive for rapid development.
Torvalds also pointed out the pressure created by automated crawlers, which not only indiscriminately extract code, but also generate low-quality reports for kernel maintainers to deal with.
He expressed frustration with links leading to irrelevant or misleading content. During the discussion, he pointed out that he rarely responds to emails, although he usually reads the messages sent to him.
The conversation then expanded to include Rust’s integration into core, the evolution of its role over two decades, and occasional friction with maintainers resistant to adopting new languages.
Asked about layoffs in software engineering attributed to AI, Torvalds called the problem complex and suggested that meaningful results might only be clear in a few years.
“It’s a complex question… AI is just one tool, in the same way that compilers free people from writing assembly code by hand and increase productivity enormously, but have not made programmers disappear,” Torvalds said.
He suggested that the industry could eventually standardize AI and treat it as routine infrastructure used alongside cloud hosting environments and cloud storage systems rather than a topic that attracts constant attention.
His comments imply that software development could divide into exploratory workflows and deeply constrained production pipelines.
The expectation is that critical code can remain tightly controlled, while rapid experimentation gains ground elsewhere.
Long-term adoption may depend on whether automated systems provide consistent accuracy rather than superficial output.
Via the register
Follow TechRadar on Google News And add us as your favorite source to get our news, reviews and expert opinions in your feeds. Make sure to click the Follow button!
And of course you can too follow TechRadar on TikTok for news, reviews, unboxings in video form and receive regular updates from us on WhatsApp Also.
