[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / s / t / u / v / vg / vm / vmg / vr / vrpg / vst / w / wg] [i / ic] [r9k / s4s / vip] [cm / hm / lgbt / y] [3 / aco / adv / an / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / pw / qst / sci / soc / sp / tg / toy / trv / tv / vp / vt / wsg / wsr / x / xs] [Settings] [Search] [Mobile] [Home]
Board
Settings Mobile Home
/g/ - Technology


Thread archived.
You cannot reply anymore.


[Advertise on 4chan]


File: apu-pillow-cry.jpg (63 KB, 736x777)
63 KB JPG
You guys told me AI progress would plateau soon... I just want to wake up from this nightmare and go back to tech being comfy.
>>
>>109770832
low iq luddite browns told you that.
>>
>>109770832
NEVER LISTEN TO LUDDITES AND/OR SNAILCATS
>>
First non bot post ITT.

What progress, OP?
>>
>>109770832
it did plateau with gpt4.5
Nothing has gotten better since. Now they are only scaling and putting more servers into the room.
Rest is marketing. Don't believe the marketing.
>>
>>109770850
Vibecoding has undeniably become passable for most normalfags' standards. It basically killed Github at this point.
>>
>>109770859
My job was not that hard, sure sometimes there were instances where I needed to think hard but most of the time, it was to take what client said, and pop that up on the screen.
AI is getting surprisingly good at this. I don't know man. It looks like it's seriously coming for my job.
>>
>>109770850
>>109770859
Explain the math stuff then.

>>109770859
Even if that were true, programming is super gay now and forever.
>>
>>109770850
Nta, but I want more jobs back, I want comfy wfh php positions back, I want to be able to get a degree and be sure that I will get a job without being 1% the most hardcore type of person in the uni, I want to be interviewed by white people, I want people to ask me about actual problems, my experience, not about leetcode, unless it's some google level corporation, no hr bs, no ghosting jobs, and no keyword filters that would waste months of my life to figure out whether it's my resume and me being genuinely bad or my interview being not parses correctly by their scanner/having a bad word in it
>>
>>109770887
>AI is getting surprisingly good at this.
no. It only FEELS LIKE it's good at this. That is the whole fucking goal of it. It should only make you FEEL good. But in fact it is not good.
It's just random words on your screen from a weighted random number generator
>>
>>109770890
>math stuff
stolen from two mathemathicians that actually made the breakthrough as their convos were part of training data
>>
>>109770890
>Explain the math stuff then.
Just read the articles, dummy.
Nothing NEW was solved/presented. All of the stuff that has been "solved" was just pushing boundaries with information that has already been inside the training data. It is just good at combining dots and searching a huge database. That's it.
Nothing new was proven. All the proofs only pushed boundaries. There was no final proof for anything.
Navier Cocks was solved (for this ONE EQUATION), because someone else actually put human labour into it and forgot to opt-out the data hoarding
>>
>>109770938
According to their press release their proof was produced in 50 hours, something no human was able to do in decades. Even if they did train their models on the input of others, that is impressive. Of course, all of this assumes they aren't lying.
>>
>>109770983
88h with the work put over a year by two human mathemathicians in the context, they literally threw 100mil to steal thunder from anthropic as one of the two was employed by them kek
>>
>>109770983
Brother. They just wrote a bot that pulled the lever automatically for 50 hours and checked if it produced the correct random numbers. That is not impressive.
The whole infrastructure and so on, from a technical point of view, is impressive, but not that it produced a result
>>
>>109770983
lmao
>>
>>109771000
burning 100 milly for 1mil prize, so on point for AI hypers
>>
>>109771024
im a snailnigger but the 1 mil is nothing compared to the billions in valuation
>>
>>109771034
more like negative billions as with the news they stole the research no phd ever is going to use their models for novel research vs some chink local model now
>>
>>109770850
OP here. Just finished my 8 hour AI furry porn goon session and I'm ready to goon again. AI ruining our lives.
>>
>>109770832
The whole "just input more data" shit already plateaued.
They're literally destroying books to find the rest of the rest of the rest of the data.
Now is the step they need to do something else like optimize the fucking code and come with new novel ways to train and use the thing, which is what is actually happening now.
>>
>>109771060
you pay them 200$ per month just so they rugpull you kek, worth every penny
>>
>>109771000
Ignoring the political drama, that is still a very impressive technical achievement. Think about how much other such context is waiting to be prompted correctly.
>>109771001
No they didn't. Even so, choosing the right numbers for a given system is part of a lot of ongoing research and is not trivial.
>>109771020
I'm suspending disbelief and speaking hypothetically of course.
>>
>>109770911
It's generative qualities are reaching human levels now. I also didn't think we would get here so quick. But the past year has been crazy.
Every single metric shows that it is getting better.It's getting better at doing everything, seriously why do the elites need us anymore.
They can just ask their bot to do it for them. No wonder we're seeing unreal levels of mass surveillance. The reason the elites are pushing for unhealthy surveillance is because they know what's coming, a revolt from the people, so they need to keep us in check.
I don't feel so good about the next 2 years.
>>
>>109770832
Nothing will ever be good again. The world ended years ago, this is Hell.
>>
>>109771304
Maybe the Mayans were actually correct, the world did end in 2012.
I just realized it was the year that Illya Sutskever won AlexNet competition using AI.
It's fucking over. We didn't realize how good 2015-2021 was.
>>
>>109771351
2016 was peak, the backlash to the woke so strong even tranny centrals like leddit buckled, imagine today we'd be able to pass a law that any LLM that can't say 'nigger' is disqualified from federal funding and needs to pass dewokization rlhf for gov use, instead we have taco shilling his next memecoin and prediction market
>>
>>109771428
>imagine today we'd be able to pass a law that any LLM that can't say 'nigger' is disqualified from federal funding and needs to pass dewokization rlhf for gov use
Nothing remotely like that happened in 2016, however it does seem like type of overeager dewoke lustration associated with the early 2025 Trump administration.
Of course I can't expect you to remember the 2016 political environment clearly since you were probably learning your alphabet back then.
>>
>>109771565
gamergate whoopped your ass, don't reply to me without the word nigger you tranny bot
>>
>>109771587
I estimate you are still in your early 20s, then.
>>
>>109771621
nigertest failed, kys tranny clanker
>>
>>109770832
It has, but these were always good at searching finite possibility spaces of well defined parameters
>>
>>109770832
I think the real feeling of getting "old" is having none of this fearmongering shit phase me anymore, not helped by the amount of shilling coming from literal bots these days
>>
>>109770850
Remember how coding needed people manually type an edit code, that is no longer needed
>>
>>109770859
Scaling is more than just compute, they are scaling training and data as well and together with compute it has shown significant progress
>>
>>109771001
>Brother. They just wrote a bot that pulled the lever automatically for 50 hours and checked if it produced the correct random numbers. That is not impressive.
That is not how LEAN proof works
>>
>>109771297
>Every single metric shows
I'm unsure about where we stand with all this AI stuff right now, but metrics have always been bullshit. If only the bean counters understood this.
>>
>>109770838
low t libtard greens warned about you
>>109771629
you first granny tourist.



[Advertise on 4chan]

Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.