@ludonaut As you're well aware if you use GrapheneOS, there are a bunch of abandoned AOSP apps which haven't had any active development for years.

Those apps are stuck in the state they were in back in Android 5 or even earlier. Many even regressed. They've had bare minimum maintenance for years which gradually ended upstream.

The issues with those apps include them not being overhauled with privacy and security improvements applied elsewhere. It includes outdated dependencies missing patches.

@ludonaut We're in the process of overhauling these apps by rewriting nearly all of the code in phases. It's becoming much higher quality code with many bugs getting fixed as part of it. That includes fixing upstream privacy and security issues which have been around for years.

Vast majority of experienced developers working at tech companies are using AI. Most of the people applying to us are using them. We have to filter through people who can pass our tests but are too reliant on AI.

@ludonaut People using AI models doesn't imply the end result is lower quality. We hired the people who we decided are capable of producing high quality code and being productive with their usual approach to development. For people who made it past our initial screening, we gave test projects and we learned a lot about what they'll produce for us from it.

All of the work they're currently doing is clearly massively raising the code quality compared to the starting point of that AOSP app code.

@GrapheneOS "additional passes of code review", you keep specifying "frontier models", why exactly do you believe this would make any difference, and how much of that donated money did you or the people you've hired spend on those death cult motherfuckers? how many tons of co2 do you think this took? how many liters of water ruined? how many lines of plagiarized code do you think? don't you love "not now/later" consent dialogs too?

actually, here, real simple, let me quote @jwz for you:

@GrapheneOS please, i'm begging you to consider that people might be sad and angry and disappointed because they've put a lot of trust in you that they feel is being betrayed. it is actually a betrayal because how are people supposed to trust their operating system, or any computers, now that it's all top to bottom compromised by torment nexus chatbots that make you feel good

@GrapheneOS
i for sure feel like a huge idiot now for ever recommending grapheneos to others, and i really don't know what i'm supposed to do now about my privacy except throw my phone into a river and walk into a forest. i don't know how many more bad news i can take, i just really wish you hadnt joined the llm bozo parade

@ludonaut We spend hundred dollars per month on it and you're claiming we're destroying the world. We disagree with the premise in the first place but even if we did agree, it's such a disproportionate response.

How is it that this is such a big deal to you but you're not attacking everyone you know who eats meat, drives a car, travels by airplane and other wasteful things? Somehow we are the problem for using less resources on this than we've used for plenty of far less useful things.

@ludonaut You would probably be fine with us running a 256 core machine 24/7 doing fuzzing but yet us paying hundreds of dollars to get far more useful results is somehow us destroying the environment.

We do not understand where you're coming from or why you're so incredibly angry with us. It's not justified and does not make sense.

Many of the people attacking us travel around the world to conferences on airplanes, drive to work and probably have the carbon footprint of half our team or more.

@GrapheneOS @ludonaut me turning off my light when i go to work while a billionaire burns my lifetime carbon footprint in an hour

@GrapheneOS It has become a religious issue.

@Lightfighter @GrapheneOS yes on one side the pro AI sicko, who think it's a god who will solve everything.

and on the other side the Anti AI sicko who see it as anti Christ who will destroy everything.

With us normal people in the middle.

@skyblitz @Lightfighter @GrapheneOS I am no fan of generative AI either, but I recognize it’s a tool that is currently needed because the opponents have them too. I’d just wish those tools would not be in the hands of those shitty companies

@Dawnexa @Lightfighter @GrapheneOS don't like AI, but sadly now the genie is out the box, and we cannot put it back.

so the only "thing" we can do without allowing the bad guy to win... is to force those company to develop the next generation to be less and less "bad" for humanity and the market.

Sadly right now due to a pedophile Orange child in the white house they do anything they want without any check on them.

@skyblitz @Lightfighter @GrapheneOS I also think a proper ethical and open source AI model would go far.

I remember when OpenAI still was „Non Profit“

@GrapheneOS
hundreds of dollars per month huh... feels great i bet.

your actions matter. you don't exist in a vacuum, everything you do is political, even diet choice or mode of travel and what to compute, especially if the "you" we're talking about isn't a single person but a project. the nth times infinity digit of pi won't matter if computing it boiled off the oceans.

lamps in videogames, as the meme goes, use real electricity, and llm datacenters run quite literally on gas turbines.

@GrapheneOS sidenote, i've had to block (edit: stopped counting) unusually hateful drive-by harassers since talking to you today, guess that's what i get for trying to help

@ludonaut @GrapheneOS They CANNOT stop. That's not an opinion, it's a fact of running anything security-critical. If you genuinely believe otherwise, then by all means, join their security team for $200 a month and show us all how it's done, analyzing every line of code that lands on their desk with the same output as an AI. Until you're willing to do that, repeating the same flawed argument isn't helping anyone.

@ludonaut @GrapheneOS
Are you vegan? And if so, do you only use software made by vegans?

@ludonaut @GrapheneOS What about other things you use in your life? Are they all made by vegans?

@GrapheneOS @ludonaut People correctly associate LLMs with fascist oligarchs and obscene levels of corruption. That is not hyperbole. That is the literal genesis of the technology. THE ENTIRE POINT OF LLM CHATBOTS IS TO INFLICT HARM. I really don’t care if you limit your fascism budget to $200 a month. The correct fash budget is zero.

You embarrass yourself and what is otherwise an attractive product with your nonstop excuses and whattabouts.

Be better and just stop using it.

@beaucoup @GrapheneOS @ludonaut Fuck off back to your Reich.

@beaucoup @GrapheneOS @ludonaut @LabSpokane
Marxism explicitly rejects moralism, arguing morality obscures the material contradictions of capitalism, such as exploitation and alienation.

@beaucoup @GrapheneOS @ludonaut @LabSpokane Luddites are resurgent! And many of us are openly COMMUNIST too ✨😈

@LabSpokane @GrapheneOS @ludonaut

They CANNOT stop. That's not an opinion, it's a fact of running anything security-critical. If you genuinely believe otherwise, then by all means, join their security team for $200 a month and show us all how it's done, analyzing every line of code that lands on their desk with the same output as an AI. Until you're willing to do that, repeating the same flawed argument isn't helping anyone.

@skyblitz @GrapheneOS @ludonaut Oh hi there 3 follower Nazi asswipe from ze Fatherland.

Believe it or not, software got written ALL THE TIME prior to chatbots, so stick that in your V-1 and smoke it.

Now, fuck off back to your underground bunker.

@LabSpokane @GrapheneOS @ludonaut it's hyperbole and you look stupid parroting this line of argument.

@MisterWanko @GrapheneOS @ludonaut Back to your wanking. Blocked.

@GrapheneOS @ludonaut this is clearly just a value discrepancy that I doubt you can resolve with words. I see the fundamental values of grapheneOS as security and privacy and those of us who switched to it with that knowledge and expectation in mind will understand using "the tools of the enemy"/"unethical tools"/whatever to the end of better security/privacy. But right now llm-opposition as a primary value is in conflict with security as a primary value. Welcome to life in dystopia.

@GrapheneOS @ludonaut My general experience is terms like AI or LLM turn the brains of generally reasonable people into mush.

Much like arguing with a Trump supporter, or an NRA member it's not really an argument you can settles with reason or rational arguments.

This is about feelings, and for many humans they trump rationality. LLMs and the extraordinarily large impact they are having on the world right now (in good and bad ways) invoke a lot of them.

@GrapheneOS @ludonaut Yeah sorry but no, even running god knows how many fuzzers nonstop will not come close to the power usage of any serious LLM. And that's just talking about inference, not even training and crawling. Even so, power use is far from the only environmental concern, and environmental concerns are from the only AI concerns people have. And you know that.

@engideer @ludonaut That's incredibly far from reality.

If we were using it for much more then we could buy a bunch of hardware and run open weight Chinese models locally. The reason we're not is because it's already being done by others.

It would be a higher priority if we were the ones responsible for dealing with the overall Linux kernel, AOSP and Chromium code getting security issues patched but we aren't. It makes more sense to spend our time and money on things not already being done.

@GrapheneOS

I'm sorry that you have to put up with this, but I believe there is a far greater majority that understands what you are doing.

Your approach seems sensible.

Using LLMs to find vulnerabilities, and accepting LLM written code as long as it conforms to standards and is readable/understandable/maintainable seems like a reasonable thing to do.

You don't have to defend yourself against every attack. Mastodon is getting pretty nasty. Some you can just ignore.

@engideer @GrapheneOS @ludonaut Frontier-scale inference uses 0.31 Wh/query. A 40 watt light bulb uses the energy of 130 LLM queries per hour. A query with large context is 13x more, but that’s still extremely marginal compared to a bunch of fuzzing machines running 24/7.

If you really want to stop destruction of the environment, stopping flying and becoming a vegetarian/vegan is much more effective than yelling at @GrapheneOS

@engideer @GrapheneOS @ludonaut Inference will only get cheaper energy-wise as more and more specialized inference hardware is being made.

At any rate, there are many studies about this, e.g.: https://www.sciencedirect.com/science/article/pii/S2542435126001145

@danieldk@mastodon.social @engideer @GrapheneOS why walk when you could roll coal instead

@engideer @GrapheneOS @ludonaut This is all besides the point that LLMs can find many vulnerabilities that fuzzers and traditional static analysis tools cannot (easily) find.

@GrapheneOS @ludonaut but funny how certain asshole, immediately label you nazi because you simply state a fact that Security = AI + human.

If we put all those people with a negative IQ in orbit, we would not see the sun anymore.

Edit : corrected QI into IQ (translation mistake), thanks matt.

@skyblitz @GrapheneOS @ludonaut Unless you're referring to the concept from Chinese philosophy I think you might have meant the eugenicist "IQ" lol

@MonKaiju @GrapheneOS @ludonaut yeah you are right i will correct it, my native language we use QI (quotient intellectuelle (belgian french)) instead of IQ thanks to pointing out my mistake 👍

@GrapheneOS you don't have to engage these people. you maintain one of the most important operating systems in use today and they contribute nothing but pissing and whining because they have experienced in the past that they get the attention they crave when it manages to hook into some general zeitgeist of concern. don't feed the concern trolls.

@ludonaut frankly you should feel like an idiot for copying someone else's absolute objection to AI. jwz is not recognizing reasonable utility of AI, and that kind of lazy thinking is exactly what we don't want here.

@ludonaut @GrapheneOS I somewhat have to agree with GOS. Most of people who hate on LLM use environmentally are not vegans and the meat/dairy industry does much more damage to the environment (and is also a literal holocaust of living beings).

I don't think LLM use is good, but I also don't think most people should speak up this harshly about it when they don't do the single easiest thing for the environment (and animal rights ofc) which is just going vegan.

I hate big tech as much as the next guy, don't get me wrong, but all the leftists are focusing on LLMs right now, because it's easier to blame a corporation than do something yourself first.

@skill @GrapheneOS sorry i forgot it's impossible for two bad things to be true at the same time

@ludonaut @GrapheneOS I didn't state anywhere that it's impossible, I'm just ranting that most people prefer to hate on AI (that will still be used, because it has a lot of value for business), where as going vegan (which reduces the demand for meat, which reduces the produced meat, which directly impacts animal lives and ecological aspects) would make a more direct positive impact.

I don't think hating on AI is bad, I am simply stating that it's mostly done by people who don't have enough will to stop doing something much more harmful themselves.

communist filth must be destroyed
in a free marketplace of ideas I win
you suck

@ludonaut Your image description keeps using "Al" (a lowercase l) in place of "AI." (uppercase "I"). You didn't use AI to write it, did you?