Agree. I’ll take the advice of Steve Gibson. “I program in assembler, becoase it’s efficient to run, not to write, but it’s flawless when it works, zero dependencies, and after compiled, good luck unwinding” I’m paraphrasing lots of things I read he said.
But yeah. New is shit, old is boomer. Fuck off genz learn to code you bunch of soy sipping cock gobblers.
I have not seen any articles about people saying how good "AI number 5" is really good at converting code to ASM and making to so much faster and more memory efficient. That might be rather telling... Don't you think?
Exactly. The problem is it’s all basically script kiddies. This is how the exploits happen, chinks infiltrate code sites like GitHub and poison the code, tiny changes here and there, benign to the dipahits that can’t code, then build trust on other good code, and then slip in rogue code that no one is capable of debugging.
Xz Utils <<<<——- you are here.
This is likely rh case with millions of exploits just sitting in unchecked code.
The researchers that find it are good, can code. But they can’t look at every git repository and look for rogue shit. Maybe ai could help do that…. Doubtful
A article I read last week (maybe it was 2 weeks ago) said that it only takes around ~230-250 articles to "corrupt" an AI. Think how easy it would be if you are a "insider" on the "training team" that likely includes a lot of cheap third world idiots?
(post is archived)