Linux is the name of a computer operating system kernel and a group of systems built on it, like Ubuntu or Mint. Linux is the basic infrastructure of the Internet. It’s important. A report ju…
The most disappointing aspect of using AI on linux, to me, is that it sort of implies linux is now more comfortable with the “move fast and break things” -mindset, now caring more about “efficiency” and “speed” rather than integrity and quality.
Open source is all about moving fast. Much, much more innovations happen in the oss space than in proprietary land.
Linux is a serious project.
As such, it has serious quality control.
Torvals already allowed AI-helped commits. And Linux didn’t break over it.
Bad pull requests rutinely get thrown out. And unfortunately for thode categorically against AI as a thing that exists, AI use doesn’t automatically make every PR bad. In the right hands it’s a great tool with a lot of potential.
And if Torvalds feels Linux is a project too big to be brought down by an onslaught of bad PRs made by wannabe script kiddies, so be it.
If that does become a crippling problem, the policy can be undone as easily as it was made. If not, competent contributors are free to use AI responsibly as a tool and not a magic wand.
A hard “no AI anywhere in any way, shape or form” stance is like the ostrich digging its head into the ground. Not trying a softer approach beforehand which allows responsible use (may also mean using approved environmentally friendly setups and open-weight models) is a bad idea for everyone involved.
Technology moves forward and we as individuals can choose to move with it or not. That being said, the foss moral imperative is: democratize any tech you can. This is especially true of when it’s a dangerous technology, which LLM-powered chatbots are.
If you instead move fast to burrow your head like an ostrich and deny the reality of the predator growing right beside you, the little lion cub that play hunted you just a while ago may very well soon grow into the lion that will devour you whole.
I only recently installed Linux for the first time on my HTPC last year and it was flawless at first. Last couple months I’ve noticed a handful of bugs and system hangs that never happened before. Anecdotal, I know, but it worries me as a novice. I built my home server on this machine specifically to avoid microslop…
Last couple months I’ve noticed a handful of bugs and system hangs that never happened before.
I get random blackscreens with my OcuLink-connected Radeon GPU on the latest kernels…anecdotal evidence, too, but my slopcode senses are tingling. It worked really well on 6.x kernels last year.
Software will probably get worse, but very slowly. One little stab at a time until it becomes very noticeable.
I find this annoying where every regression now gets associated with “AI slop”. I agree that genuine slop is bad. But Radeon regressions were pretty damn common pre-AI-slop too. My 6600M, my Vega iGPU and others all had regressions. And I’ve seen this pattern with everything else lately too, where even bugs that have existed for over a decade are associated with LLMs. Pointing things out specifically is important in these scenarios IMO. “This LLM-generated and claimed-to-be-human-reviewed patch caused this regression.” That’s much more useful than “slopcode senses” that people seem to be triggering in projects that have existed for so long. (I do get where you’re coming from though…)
The most disappointing aspect of using AI on linux, to me, is that it sort of implies linux is now more comfortable with the “move fast and break things” -mindset, now caring more about “efficiency” and “speed” rather than integrity and quality.
Open source is all about moving fast. Much, much more innovations happen in the oss space than in proprietary land.
Linux is a serious project.
As such, it has serious quality control.
Torvals already allowed AI-helped commits. And Linux didn’t break over it.
Bad pull requests rutinely get thrown out. And unfortunately for thode categorically against AI as a thing that exists, AI use doesn’t automatically make every PR bad. In the right hands it’s a great tool with a lot of potential.
And if Torvalds feels Linux is a project too big to be brought down by an onslaught of bad PRs made by wannabe script kiddies, so be it.
If that does become a crippling problem, the policy can be undone as easily as it was made. If not, competent contributors are free to use AI responsibly as a tool and not a magic wand.
A hard “no AI anywhere in any way, shape or form” stance is like the ostrich digging its head into the ground. Not trying a softer approach beforehand which allows responsible use (may also mean using approved environmentally friendly setups and open-weight models) is a bad idea for everyone involved.
Technology moves forward and we as individuals can choose to move with it or not. That being said, the foss moral imperative is: democratize any tech you can. This is especially true of when it’s a dangerous technology, which LLM-powered chatbots are.
If you instead move fast to burrow your head like an ostrich and deny the reality of the predator growing right beside you, the little lion cub that play hunted you just a while ago may very well soon grow into the lion that will devour you whole.
I only recently installed Linux for the first time on my HTPC last year and it was flawless at first. Last couple months I’ve noticed a handful of bugs and system hangs that never happened before. Anecdotal, I know, but it worries me as a novice. I built my home server on this machine specifically to avoid microslop…
I get random blackscreens with my OcuLink-connected Radeon GPU on the latest kernels…anecdotal evidence, too, but my slopcode senses are tingling. It worked really well on 6.x kernels last year.
Software will probably get worse, but very slowly. One little stab at a time until it becomes very noticeable.
I find this annoying where every regression now gets associated with “AI slop”. I agree that genuine slop is bad. But Radeon regressions were pretty damn common pre-AI-slop too. My 6600M, my Vega iGPU and others all had regressions. And I’ve seen this pattern with everything else lately too, where even bugs that have existed for over a decade are associated with LLMs. Pointing things out specifically is important in these scenarios IMO. “This LLM-generated and claimed-to-be-human-reviewed patch caused this regression.” That’s much more useful than “slopcode senses” that people seem to be triggering in projects that have existed for so long. (I do get where you’re coming from though…)
That’s really sad actually, i didn’t think of that.