AI WON and there is nothing you can do but cope
>>108774801Uh-huh, and how many of those bugs were introduced by the aforementioned AI?
I threw in the codecel towel, swallowed my pride and bought a Codex and Claude Code pro subscription, just 20 bucks so we are talking about skipping 8 time a coffee to goNow im getting 10x more shit done, feels good to not be a snailcat anymore
and that's why firefox is the buggiest POS it's ever been?
>>108774801so that mean firefox will finally get better than chrome/chromium right?
>>108774801> it's full of bugs nowcode monkeys and AI jeets are equally useless
>>108774801Have you seen those security bugs? Just a conditions that'll never happen irl but AI blindly fixed them as a security bugs
>>108774842>just 20 bucksFor now.
>>108774880no, chromes' security architecture is far superior and properly sandboxed.https://madaidans-insecurities.github.io/firefox-chromium.html
Now go through all 2 million bugzilla reports.
>>108774923>last updated: 2022
>>108774923Firefox is especially awful on Linux. I switched to a chromium browser because of this.>>108774936Still hasn't been fixed in modern updates
>>108774801>AI adds 1000 bugs>AI fixes 400 bugs
>>108774914>For now.for ever since there is tons of competition
>>108774960>there is tons of competitionFor now, AI will crash and that's the end of that.Just like the dot com boom, the internet went to shit soon after.
>>108774942>i am going to pull shit out of my ass
>>108774974>For now,even if models won't improve at all <impossible> they will get /2 in price every 2 years
>>108774975>butthurt troonyfox user
>>108774988They're still running these things at a loss, anon. They're eventually going to make you pay per token.
>>108774988They're going UP in price anon, and the entire hardware market is going up because of them
>>108774801What about memory leak bug fixes?
>>108774989https://bugzilla.mozilla.org/show_bug.cgi?id=1129492clicking a random link from your garbage outdated shit link shows how full of shit you are and why you have not updated your garbage in years. most of this have been fixed but because you are a googletron you want to keep shitting on basedfox. exposed.
>>108775013>Took 7 years to "fix"Wow such an amazing browser
>>108774997>They're still running these things at a loss,not true
>>108775033https://bugzilla.mozilla.org/show_bug.cgi?id=1653444https://bugzilla.mozilla.org/show_bug.cgi?id=1530770https://bugzilla.mozilla.org/show_bug.cgi?id=1302711https://bugzilla.mozilla.org/show_bug.cgi?id=1129492oh wow look every single complain has been fixed!EVERY SINGLE COMPLAINT HAS BEEN FIXEDUpdate your garbage hit piece you coward googletron piece of shithttps://www.youtube.com/watch?v=o_m1J4KzhXg
>>108775059>All bugs takes years to fixThat's years that people are exposed through retarded vulnerabilities that should have never existed. Face it troonyfox user, Chromium is the future.
>>108774801You think LLMs are good at coding? You're wrong.Whenever they happen to generate code that is passable it was code that was previously written by a human. They may be able to translate it to another programming language - they're quite good at that (they're language models after all), but they are truly awful at generating anything they didn't see and remembered during their training.Btw. those fuckers are aggressively stealing private code. And I don't mean from private but published online repos, no, I mean from your local machine if you happen to have some LLM client running on it.
>>108775033I think they have a policy that if no one is complaining then they don't do anything about the bug.I encountered a bug once writing some wasm library, but didn't know if it was actually a bug or expected behavior at the time, went to their chat - they pointed me to the bug report that was a couple years old and then fixed it after a few days.
>>108775068update your garbage hit piece website goolag troon. all of the bugs have been fixed
>>108775134>You think LLMs are good at coding? You're wrong.Whenever they happen to generate code that is passable it was code that was previously written by a human. They may be able to translate it to another programming language - they're quite good at that (they're language models after all), but they are truly awful at generating anything they didn't see and remembered during their training.Btw. those fuckers are aggressively stealing private code. And I don't mean from private but published online repos, no, I mean from your local machine if you happen to have some LLM client running on it.Cope Luddite
>>108775171ai post
>no snailcat
>>108775134Cloudflare just (literally like an hour ago) laid off 20% of their workforce citing efficiencies in using AI agents:https://blog.cloudflare.com/building-for-the-future/Everyone has their head in the sand or is lying to themselve. It's literally happening like the AI CEOs said it would and no one seems to be able to admit it.I don't like it because I spent 4 years getting a CS degree that will probably mean nothing in 5 years. But running from reality is just stupid.
I hope f95 integrates claude mythos to check for malware
>>108774814If they apply to earlier versions of FF pre-AI boom, then none.And that is probably the case with most, so likely very few. You can go and check them if you want to make a case. But I'd bet a lot that these bugs also applied to older versions, making your point moot if true.
>>108775202They're shedding excess employees after COVID overhiring and using AI as an excuse.Btw. this overhiring had nothing to do with increased demand for code or anything like that. It was about financial incentives from both the government and financial institutions who were trying to decrease the impact of the pandemic on the economy - not hiring thousands useless employees at the time would be like passing on billions of free gov money flying by your nose.
>>108775202nope that's fake ai can't do anything, cope.
>>108775202https://h1bdata.info/index.php?em=cloudflare+inc&job=&city=&year=202599 jobs that americans could have btw. this is only ones that they imported to america, they have like 2000 overseas/contracted jobs too. also this is only for 2025, h1bs last 3 years and cna be extended for another 3, so you'd have to add 2025+2024+2023+2022+2021+2020 which is like 600 jobs at cloudflare are taken by H1Bs
>>108775240What covid hiring? :^)>As of December 31, 2025, we had 5,156 full-time employees, including 2,452 employees located outside of the United States. We also engage contractors and consultants. >As of December 31, 2024, we had 4,263 full-time employees, including 1,896 employees located outside of the United States. We also engage contractors and consultants. >As of December 31, 2023, we had 3,682 full-time employees, including 1,592 employees located outside of the United States. We also engage contractors and consultants. >As of December 31, 2022, we had approximately 3,217 full-time employees, including approximately 1,332 employees located outside of the United States. We also engage contractors and consultants>As of December 31, 2021, we had approximately 2440 full-time employees, including approximately 975 employees located outside of the United States. We also engage contractors and consultants>As of December 31, 2020, we had 1,788 full-time employees, including 666 employees located outside of the United States. We also engage contractors and consultants.>As of December 31, 2019, we had 1,270 full-time employees, including 439 employees located outside of the United States. We also engage contractors and consultantsall information taken from annual 10-K filings https://cloudflare.net/financials/sec-filings/default.aspx
I've never hard a problem with any of the browsers I've been using.
>>108774914Yeah, in a few years its going to be $1/month for unlimited Opus since the models will be hardcoded into Asic chips. Cope and seethe.
>>108774997>>108775001>>108775036People need to start investing into local hardware fast then. (We should have been doing that years ago in hindsight). If you can run the models locally yourself, you're fine for the most part. It's the people completely and utterly dependent on apis that might be fucked. I don't believe the bubble is going to last either, but I'm not convinced con API AI. It's just going to evaporate or disappear into nothingness. All it really means is that the big players that can afford to take the financial aid will just buy out the failing competition. It's a bit too early to say for certain whether or not they will jack up prices to the Moon, prices will fall, or neither.
>>108775295Every time one of these tech layoffs happens (and it's basically every week it feels like) people always reflexively reply with that "they're just correcting from COVID!!!!" line. It's such utter cope. I don't like the way the world is heading either, but ignoring it with your head in the sand is not the right approach.
>>108775618Local does seem like a more future proof approach for the little guy. The problem is it's relatively expensive and you need a good GPU and lots of RAM, right? What's the worst GPU that can run something decent? The only GPU I have is in an ancient broken prebuilt from like 2015.
>>108775774Relevant backends like llama.cpp and ollama are for the most part, pretty hardware agnostic (support for AMD, METAL/Apple Silicon, and of course CUDA, etc.). Your main limitation is the amount of vram and system ram you have. Vram determines the size of the model you can actually run and whether or not you can even load it onto the GPU in the first place. For my personal use in workbooks locally, I don't trust anything lower than the ~20B range for vibecoding. (A QK_4_M quant of GPT-OSS 120B runs lightning fast on my MacBook mostly due to it being a moe). As for what's the worst you can run that depends on what you want to do. If you just want to do general purpose stuff then you can get away with going as low as 14b but I wouldn't go any lower, which means you realistically would want a GPU that is in the 16 GB range minimum, along with adequate system ram. 20b or higher, you're going to need a 5090 tier GPU in order to actually run it and to get an acceptable amount of tokens per second. You can technically get away with having a weaker and lower memory GPU with llama.cpp specifically by setting up the local server to essentially have the "experts" sit completely on the RAM while the rest of the model is on system ram, but then your tokens per second fall off a cliff, especially at longer contexts. If you want less headache actually getting it to work And being able to use models that are actually worth a damn you going to have to be comfortable shelling out some coins. My MacBook with 128GB ram along with 4tb of storage cost me just north of 6 grand but it's more than worth it if it means I can run these types of things locally anytime I want it anywhere I want. I can afford a proper desktop rig that's actually powerful, but then I'm constrained to only doing these types of things in my room, and I'm out of the house and mobile pretty frequently both due to my career and responsibilities.
>muh AIIt's literally starting to sound like shitcoins at this point. It's so fake and gay I don't understand it.
>>108775899>t. snailcat
>>1087758992.66T market cap on crypto btw
>>108775295>4.06x employees in 6 yearsHueg growth. Were they bloatmaxxing or smth?
>>108775295>>108775752>size of the company tripled during COVID>what COVID hiring?Your shilling is obvious. Don't know why I'm even responding since there are no real people itt.
>>108775899Civilization runs on slavery (always). Those who are privileged enough to not be slaves are scamming each other to get fruits of their (slave) labor.
>>108775929>t. indian>>108775952pay up the taxes, IRS will take it
>>108774842What a meme reply. Painful to read.
Marketing image, probs put their by the devs. AI worse than people think. But when was the last time humans lost a war? We've been down to very small numbers during extinction events. AI will be the morons but not all of us.
ITT: coping luddites
>>108776600T. arresting libtard
>>108776600ESL
>>108774801AI is GOD.
>the only ''people'' who need FF are darpa's tor and google's anti-lawsuit sad but true
>muh jewgle sandboxingtell that to my 3000 open tabs
>>108774801Did anyone check if anything changed was actually a bug?
>>108776600boop
>>108775165>all of the bugs have been fixedlemao 4 bugs out of 100000000, firefox is in iths death bed, it's the worst it has ever been right now.
>>108775618you should update your OS bruv
>>108778853I checked few patches randomly and it's mostly edge-case bugs that are nearly impossible to attack from outside I think, some hypothetical invalid pointer calls and so. But as I said, I haven't checked all, I don't have so much free time
>>108775221Just say "I don't know" next time.
>>108774801Did Anthropic RED partnership only last one month or what?
>>108779719they run out of gpus
>>108774960>>108775584Has any AI company even turned a profit?Shit is going to be much more expensive as soon as the VC money dries up.
>>108775202>Everyone has their head in the sand or is lying to themselve. It's literally happening like the AI CEOs said it would and no one seems to be able to admit it.At least I work in a slow moving industry(Healthcare and insurance) so my job is safe for a couple more years. But we're definitely hitting a place where it's able to do like 80% of coding.
>>108774842100 tokens have been credited to your account.
>>108774801Firefox has been crashing so fucking much for me lately. Is this shit why?
>>108774801great, now fix all the mem leaks.
>>108774842>Claude Code>$20>xhe doesn't know
>>108779781>Has any AI company even turned a profit?since when did start ups have to turn a profit during growing phase ? do you know anything about economics
>>108779781I'd like to remind you of what happened with Amazon. They lost money for a very long time and did that to grow until investors really got annoyed, and then they finally started making money. I can see AI companies pushing for datacenters and training models for as long as it's tolerated. When investors finally getg sick of it, they'll stop wasting money and focus on making money. That'll allow them to push models as far as they can and get as much infrastructure in place for the long-term. Then they only have to pay to keep the servers on.
>>108774842I wonder who pays these Indians to shill this trash here.
>>108780158>Indians to shill this trash here.indians are the first to get wiped out by ai chud get a fresher cope
>>108776600WHAT!! Remember, snail-chads will holocaust your indian ass, beware, and besides we don't care about your jewish agenda!!!
>>1087748011000% of 0 is 0
>>108774801POG
>>108779451All of the bugs have been fixed. That is the reason you don't want to update your garbage hit piece Globohomo disinfo website. Goolag troon update your website.
>>108780581:DDDDD
>>108774801did any one correct the /g/eets who think you can't let AI find bugs then fix them yourself.i had to stop by like the 10th comment (daily IQ loss limit reached).
>>108779815No, it's pretty stable. I guess you got faulty RAM, run Memtest86+, I had this too, replacing faulty RAM memory sticks solved this
>>108774801>We live on a rock orbiting a massive fusion reactor which can projectile vomit charged particles that have fucked with the grid before.>Grid would've been cooked over a decade ago but [un]fortunately the CME missed us.Maybe the Euros have a decent electrical grid and shit, but amerimutts are too jewish and retarded to update their grid. All it takes is for the Sun to sneeze and we're screwed.
>>108774801>AI WONby forfeit. Mozilla was supposed to rewrite their browser in Rust to not have all those vulnerabilities but they never did.
>>108776600the Rust language is for luddites?
Firefox legitimately is so much fucking better and I have my name on ti
>>108775202>if they said that's the reason, that's reasonthey have no reason to lie about, right?
i can run 1 bit on my thinkpad just fine, hook him up to wiki and he's smart as fuck, i am not afraid of some boogeyman price hike, my ~4gb file on my disk has so much fucking knowledge i'm amazed
>>108774801I admit that AI can hack now, but it still can't program for shit. This means that good programmers are going to be even more needed, because of AI hacks.Thanks AI.
>>108783484It can. I literally ported entire IDEs to novel operating systems with it. It's good enough. It sucks at writing an SEPolicy though. But how many people do you know who can write an SEpolicy?
>>108783487>It can.False.>It's good enough.False.I don't know what's SEpolicy, I guess it's some convulted configuration shit.
>>108783497So just be honest with me right now, are you in the industry or not?
>>108783501yeah call up zuck he can verify im in the scene
>>108783501yes
>>108783512https://docs.redhat.com/en/documentation/red_hat_enterprise_linux/7/html/selinux_users_and_administrators_guide/chap-security-enhanced_linux-the_sepolicy_suite
>>108774801>AI WON >and there is nothing you can do but copeI could kill bmyself.
>>108774801going to doubt AI slaves until it starts taking all the white collar jobs.
>>108783476You're not doing shit.
we need developpers, i hardly see a world were trains/cars/spaceships/flights are coded by AI (not 100% reliable)at the same time most of programmers are going to be replaced because with LLMS you can shit out any end-user program without any problem depending of the size of the project but for securities problems you'll still need someone.coders are now competing against a tiny portion of no-coders who shit out apps like there's no tomorrowno ads/no extra-telemetry/tokens used maybe few dollars instead of paying subs
>>108774801It did. I am seriously looking into changing profession, because us, 0.1x devs, won't survive that.
>>108774842>20 bucks >8 coffees damn, even in my slavic shithole 20 american USD made in US will get you at best 5-6 coffees from regular coffee shops
>>108774801you should kys
>>108774801holy shit!
>>108780076OpenAI, Anthropic, Google, Microsoft and friends are not startups, no matter how you look at them.>>108780144>When investors finally getg sick of it, they'll stop wasting money and focus on making moneyAnd how do you expect them to do that short of massively increasing prices?>Then they only have to pay to keep the servers on.You almost make it seem like that is the cheap part.These are GPU-centric datacenters. The power costs are orders of magnitude higher than the standard datacenter.Again, when the investor money ends, they're going to charge you obscene amounts.
>>108780144One thing many seem to overlook is that they will still have to use money, time and energy to train their models otherwise they'll stop being useful overtime as tech progresses beyond their training data. So it's not just paying to keep servers on
>>108774942i use it daily with librewolf and using it right now. never had an issue besides being to strict as usual but thats my fault.
>>108780144Yes and Amazon got more expensive. Used to be that Amazon was reliably the cheapest source for nearly anything. Now the "free shipping" is baked into the prices and you can often do better by shopping elsewhere.Take your pick of companies that did this—Uber, Netflix, any of them. The product gets shittier and more expensive when they switch into the extractive phase.The difference with LLMs is that in theory the enshittification *could* be absorbed by progress in the technology. I doubt it will though
>>108774842>8 coffees for 20 bucks>coffee to go for 2.5 bucks in 2026whatever you say you fucking indian
>>108783484It can't hack either, it can find things that resemble bugs. Sometimes they even are bugs!
>>108775202>https://blog.cloudflare.com/cloudflare-1111-intern-program/really makes you think
>>108785817say this in the wrong hood and you'll get your wig split by someones agent
>>108785827>Bengaluru, India
>>108774801AI makes mistakes, it's "imprecise computing".As long you understand that, i see no problems.
>>108785927>implying humans don't do mistakes kek the cope is unreal
>>108785962I'm not, but we design systems around humans making mistakes.The same has to be applied to AI, or you will end with a flood of shit.
