![[BunsenLabs Logo]](/img/bl.svg)
You are not logged in.
- Topics: Active | Unanswered
#101 Yesterday 22:01:17
- Martin
- Member
- From: Stockholm, Sweden
- Registered: 2015-10-01
- Posts: 842
- Website
Re: AI discussions
"Problems worthy of attack
prove their worth by hitting back."
Piet Hein
Offline
#102 Today 02:28:35
- johnraff
- nullglob
- From: Nagoya, Japan
- Registered: 2015-09-09
- Posts: 13,471
- Website
Re: AI discussions
^excellent, thanks.
The instinct today is to let AI do the entry-level work because it is faster and cheaper. But some of that work is not overhead to be eliminated. It is the training apparatus of your future senior staff, and you should protect it the way you’d protect any other piece of critical infrastructure.
...elevator in the Brain Hotel, broken down but just as well...
( a boring Japan blog (currently paused), now on Bluesky and Mastodon, there's also some GitStuff )
Offline
#103 Today 02:48:43
- johnraff
- nullglob
- From: Nagoya, Japan
- Registered: 2015-09-09
- Posts: 13,471
- Website
Re: AI discussions
It cuts both ways, useful at times, and flat out wrong at times, this is why I shudder when I think of heavy AI usage in the kernel, 30 million+ lines of code it just doesn't seem possible that someone is actually auditing the gargantuan amount of patches AI is generating for it.
Well, all those 30 million lines were originally written - and audited - by human beings. Nobody checked over the whole lot at once.
It seems that the quality of AI generated bug reports might have improved over the last year or so, but anyway the bottomline is that each patch has to be signed off by a human developer. They take personal responsibility so there is some motivation to do that auditing seriously.
Torvalds and the maintainers are acknowledging reality: developers are going to use AI tools to code faster, and trying to ban them is like trying to ban a specific brand of keyboard.
The bottom line is, if the code is good, then it's good. If it's hallucinatory AI slop that breaks the kernel, the human who clicked "submit" is the one who will have to answer to Linus Torvalds. In the open-source world, that's about as strong a deterrent as you can get.
https://www.tomshardware.com/software/l … -agreement
https://byteiota.com/torvalds-ai-linux- … velopment/
https://docs.kernel.org/process/coding-assistants.html
PS another thing is that AI is turning up many security vulnerabilities that have been undetected for decades. That's a plus, I'd say.
...elevator in the Brain Hotel, broken down but just as well...
( a boring Japan blog (currently paused), now on Bluesky and Mastodon, there's also some GitStuff )
Offline