Obligatory xkcd reference: https://xkcd.com/378/

It's disappointing to see Linus pull out the tired "compiler" metaphor. AI agents are obviously not compilers, or we'd be committing prompts to Git. I think anyone could think of a dozen ways they're unlike any other tool a programmer has access to, so if he wants to convince anyone who doesn't already agree with him, he needs to make an argument for why they're like the other tools, not start from the conclusion that they're like them and ridicule people for treating them like they're not.

But Linus stands to financially benefit from his stance on AI. The current AI boom runs on Linux servers. The Linux Foundation is in deep with AI projects and its current platinum members are some of the biggest AI promoters. It would be very surprising and insanely self-destructive for him to take a principled negative stance on AI like we see from some other free software projects, but that also means you can't read a lot into his position. If you think money and power can influence beliefs at all, there's so much of it at stake right now (or at least there's the perception that there is) that you can't really trust people like "the guy who wrote the operating system that everyone uses for inference."

Some of the Linux-adjacent projects I've really come to respect in the years before this all became so serious are the ones bringing Linux to new unusual platforms, like Asahi and Nura (nee postmarketOS,) both of which have strong public anti-AI stances. My IdeaPad Duet Chromebooks running Nura are some of the most enjoyable Linux devices I've used, so as much as I let any talented developers influence my position on AI, I'd listen to them before I listen to Linus.