it is getting there, and not so slowly though. The remaining problem is that it's still just a tool. Telling a random dev "make zig faster in a one shot PR" isn't going to give good results either.
In the past, OSS projects were self-selective because you needed to be able to make working code, and if you did, you probably also reasonably did the right things as you spent years learning this, and have some sort of reasoning behind your feature, need, etc.
Today, even if the LLM was perfect and could reason well, it still does the bidding of the prompter - and you no longer have self-selection. Heck, it'll be difficult for zig devs to decide what's actually made by an LLM or a human anyway, I'm sure there's already LLM generated code in there - but at least these [human submiters] still need to be reasonably good at code.
I wonder if we'll end up with "only human with trusted badge of honor" can commit, and/or "LLMs now reason well enough to tell you: 'no, f off, this feature, plan, idea is garbage I'm not generating it" hehe.
When i was a kid, my dad had a Mac with the A/V PAL-SECAM cards. Hooked up a make-shift copper wire antenna and wrote a decoder with the free codewarrior cd folks gave me at Paris' Mac convention (we were 12 and crazy I guess). Good opportunity to learn powerplant and c/c++.
I ended up brute forcing most of it as I did not really understand what I was doing, but it turns out, with enough time, you get things going.
Wish the pages were still up, I lost that software long ago, and I'm sure my code was garbage (not that its much better today, but at least I can blame Claude..) and fun to read.
- verifying isn't asking "is it correct?"
- verifying is "run requests.get, does it return blah or no?'
just like with humans but usually for different reasons and with slightly different types of failures.
The interesting part perhaps, is that verifying pretty much always involves code, and code is great pre-compacted context for humans and machines alike.
Ever tried to get LLM to do a visual thing? why is the couch at the wrong spot with a weird color?
if you make the LLM write a program that generate the image (eg game engine picture, or 3d render), you can enforce the rules by code it can also make for you - now the couch color uses an hex code and its placed at the right coordinates, every time.
google even has specially signed fw that let you root the device and unlock anything that doesn't rely on the passcode. secureboot passing and all. i can't imagine that the nsa doesnt have them. after that you just gotta crack the usually very simple passcode. wouldny be surprised if thats what cellrite has lol.
you seem to believe that llm are a neutral engine with bias applied. its not the case. the majority of the bias is in the model training data itself.
just like humans, actually. fe: grow up in a world where chopping one of peoples finger off every decade is normal and happens to everyone.. and most will think its fine and that its how you keep gods calm and some crazy stuff like that.
right now, news, reddit, Wikipedia, etc. have a strong authoritarian and progressive bias, so do the models, and a lot of humans who consume daily news, tiktoks, instagrams.
I think that there's also inherent trust in "hardware security" but as we all know its all just hardcoded software at the end of the day, and complexity will bring bugs more frequently.
to be fair, most of MTE's benefit is realized by having enough users running your apps with MRE enabled, rather than having it everywhere.
This is because MTE facilitate finding memory bugs and fixing them - but also consumes (physical!) space and power. If enough folks run it with, say Chrome, you get to find and fix most of its memory bugs and it benefits everyone else (minus the drawbacks, since everyone else has MTE off or not present).
trade offs, basically. At least on pixel you can decide on your own
Dan Kaminski popularized this in 2007-8 or so. Not that it didn't exist here and there, but he made the perhaps first public version of a dns tunnel (ozyman). he inspired iodine and others and was a fairly well known guy.
Dan passed away in 2021, rip.
if you search for it its hard to find. his blog is down (hea dead...), and many companies and people talked about it on his behalf to drive traffic (hi duo sec..), so you can see the internet forget, rediscover, and rewrite some history even in a few years.
a lot of the western world learns only speaks about ww2 (let alone ww1, americans civil war, etc.).
there has been countless western and non western wars with slightly different patterns and a taste of "winner writes history".
one i find interesting is the french revolution. its also fairly recent, but not as tampered with as ww2 history. for example, there still are records of how terrible and cruel the revolutionaries were, how everyone was a royalist that needed to die and how the populace started to be ready to revolt - again - right after the change of power. thankfully, things eventually calmed down - as they were cruel, but not dumb.
either way I'd basically recommend expending the reading curriculum a bit.