Now ... the title is catchy and I will state upfront that Ginux is not about to replace Linux - at the least not anytime soon. But probably also not in the future, as the Linux kernel having a team of real human behind itself, is IMO a stronger foundation than a ChatGPT-created Operating System, such as Ginux.
Anyway - the article is still quite interesting.
"In 11 days, I built a new Linux-compatible operating system kernel that boots Rocky Linux 10 with systemd, all using Human-Directed Agentic Engineering (HDAE)."
This is quite interesting though. Imagine if it would take 11
days only, to replace all Top 500 supercomputers running Linux
with Ginux or anything similar. Again, not saying this is likely
right now but I think sooner or later this may happen. Or at the
least "real" competition.
Of course Ginux right now is nowhere an equivalent:
"Ginux runs only in a QEMU virtual machine, with no physical hardware profile"
And solving the hardware problem is big; HaikuOS still struggles
here, ReactOS even more.
But this is also a finite problem set; soone or later AI slop agents
will have figured that out. And then it will become more interesting.
Personally I have no interest in AI-linux or AI-operating-systems, but you can see that omarchy slop built by DHH is popular (see
distrowatch, it is on rank #11 or something like that; yes, that is
only a partial view, but TIOBE is also only a partial view, but still not 100% rubbish, just ... 80% rubbish 20% interesting).
Of course peer review by humans is more important than self-promo description by the human creator of Ginux, but I think sooner or later that'll also be solved.
Anyway - the article is still quite interesting.
"In 11 days, I built a new Linux-compatible operating system kernel that boots Rocky Linux 10 with systemd, all using Human-Directed Agentic Engineering (HDAE)."
This is quite interesting though. Imagine if it would take 11 days only, to replace all Top 500 supercomputers running Linux with Ginux or anything similar. Again, not saying this is likely right now but I think sooner or later this may happen. Or at the least "real" competition.
Of course Ginux right now is nowhere an equivalent:
"Ginux runs only in a QEMU virtual machine, with no physical hardware profile"
And solving the hardware problem is big; HaikuOS still struggles here, ReactOS even more.
But this is also a finite problem set; soone or later AI slop agents will have figured that out. And then it will become more interesting.
Personally I have no interest in AI-linux or AI-operating-systems, but you can see that omarchy slop built by DHH is popular (see distrowatch, it is on rank #11 or something like that; yes, that is only a partial view, but TIOBE is also only a partial view, but still not 100% rubbish, just ... 80% rubbish 20% interesting).
Of course peer review by humans is more important than self-promo description by the human creator of Ginux, but I think sooner or later that'll also be solved.