i'm just glad i was able to make some software bucks before OpenAI knife fucked the job market.

my sister laughed at me today.
this sucks.

>>103593650
A pajeet using AI will replace you soon enough

>>103593661
why would they even need to hire a pajeet tho?
they'd just get in the way.

>>103593661
pajeets are fucked too, voice AI will replace all the call center jobs, its fucking over for everyone

>>103593650
its gonna be a technocracy.

0.1% "decider" class
0.9% "supervisor" class
99% cannon fodder

>>103593650
buy an ad

>>103593650
Let's hypothetically assume AI can write actual useful software. Not code. Software. This thing would only be able to program concepts that it has seen, and never be able to create anything new. If you job entails monkey works like gluing libraries together for shitware, start a new career in plumbing. If your job is about implementing new products, mathematics, computer graphics, engineering, simulation, you are good.

>>103593650
kek i hate programming too

>>103594668
It got a good score in Fronteir Math which is designed to have questions that are not well published in literature. Thus it is solving problems it has not seen.

>>103594722
There is a difference between solving a niche math problem vs designing an entire software stack that relies on applying the concepts i listed earlier.

>>103593650
The timing is perfect. The world, in its infinite wisdom (or absurdity), has decided that Chuck's Seed & Feed must now be Sneed’s Feed & Seed, because the cyclical nature of life demands that the narrative be pushed forward, ever closer to the edge of a great and inevitable punchline.

File: awef3f.png (46 KB, 197x262)

46 KB PNG

AI will never take away my joy of learning!! >-< i'll program because i want to! because it's fun!

>>103594742
Yes, the math problem is much harder

>>103594760
irrelevant, because it doesn't implement any useful software, which is the entire point of the fucking thread

>>103594775
Hush brainlet, you've been deprecated

>>103594742
The contention was that it would only be able to program concepts that it has seen. The Frontier Math result shows that it can solve novel problems in some cases. So it's a big leap.

File: SAAR Bubble.png (85 KB, 338x208)

85 KB PNG

>>103594801
>Hush brainlet, you've been deprecated

>>103594668
Where does this dumb idea come from that the LLMs can only do stuff they've "already seen?" Clearly you have no idea how this shit works.
This is coming from an LLM skeptic that doubts they can build software. But I believe we should at least argue with sense

>>103594844
>Clearly you have no idea how this shit works
t. someone who clearly has no idea how this works

>>103594752
born to sneed, forced to chuck

>>103594856
Just spend 10 seconds with an LLM and it's pretty clearly able to generate text that has never existed before. "Write me a poem from trumps perspective about how tasty big macs are" etc.
>but software is different!!!
No it isn't. Software has even less variation than spoken language.

>>103594844
LLMs is just an elaborate predictive inference machine. It has no capacity of congnition, creativity, analytical thinking, or understanding concepts. It just generates shit by rolling a loaded dice. I've been implementing/training neural networks since the 90s, and I can tell you, nothing revolutionary is happening here, other than more compute power lets a few tech bros to run models that doesn't suck as much as they did before (but they still do). Asking LLMs to write am entire shippable software product, say like a AAA game title, is like expecting a spell checker to write a whole fucking novel for you. Dickhead.

>>103593650
Its over code monkeys, should've learned something useful.

>>103594922
Wow are you an idiot or what? Your reply is completely unrelated to what I wrote. I don't believe LLMs can write software and I agree with most of what you said. What made you think I didn't, idiot?

>>103594882
>able to generate text that has never existed before
Bullshit. It just randomly strings words/concepts together it already seen during training. The randomness is heavily biased towards a particular set of outcomes, which happens to make sense for a human reading it. And of course it would, the fucking thing scraped data that was written by humans.

>Software has even less variation than spoken language.

Which makes spoken language easier to generate, because it allows more flexibility of expressing vague vocabulary to describe something.

>>103594977
>Where does this dumb idea come from that the LLMs can only do stuff they've "already seen?"
I responded to this, pissant, which makes my response 100% relevant.

>>103593650
Competitive programming is not a job. If your uncle who works at Nintendo claims it compares favorably to actual job skills, he's tryna bait you.

Yay, another fake benchmark from OpenAI. Remember when o1 was supposed to be the cream of the crop and it couldn't even edge out Sonnet 3.5?

I'm a /sci/ regular and sometimes I come here to see how this supposedly intelligent board reacts to AGI being realized

But...is everyone here in denial because they will lose their job, or are you seriously stupid and can't see what is happening?

>>103595177
which jobs are most threatened by this new development in your opinion?

>>103595177
>/mu/ has terrible music taste & is a terrible board
>/v/ has terrible video game taste & is a terrible board
>/tv/ has terrible movie/tv taste & is a terrible board
>/g/..

>>103595376
All of them? With software dev and fucking MATHEMATICIAN being possibly made irrelevant by 2026 at this point.

>>103595398
Not true, I remember /g/ being decent years back, but it just seems to have adopted the absolutely worst take on AGI and LLMs. Seriously even lobotomized subreddits get it by now, and people here don't

>>103595512
how will Ai replace plumbers or any job requiring complex physical interaction with the real world. You can't really mean all jobs. What I guess you mean is any math / coding based job?

What about network administrators and computer technicians?

>>103595531
ofcourse it can replace plumbers. Specially if the housing construction is standarized. Repairing plumbing would be more difficult but not impossible.

>>103595557
how is an AI going to fix a broken pipe?

>>103595559
first itd have to recognize the pipe which is currently possible. Then itd have to detect where the pipe fits which is also currently posdible. Then itd have to grab a wrench and level to where the pipe fits. Then itd turn the wrench remove the pipe and put a new one in place Easier said than done though.

>>103593650
The future will be zero class mobility. If you're born smart but poor, like some of you, it won't matter. You'll stay poor forever.

>>103595598
how could that be for every industry? Business men, athletes, investors etc can still change class

File: 20241221_105428.jpg (47 KB, 1054x1046)

47 KB JPG

>>103593650
Why are people posting the less interesting benchmark
This one is insane

>>103595177
/sci/ is full of dunning kruger retards just like every other board on this site

>>103595177
None of the models are capable of doing high school level math. You are actually retarded. None of them are intelligent, show reasoning or understanding. You are being scammed. You are from /sci/ so you should know how retarded these models are when you ask them basic high school level science/math questions.

>>103595512
>MATHEMATICIAN being possibly made irrelevant by 2026 at this point.
Kek keep dreaming. AI cannot do basic math and never will.

>>103595675
>None of the models are capable of doing high school level math.
Bad bait.

>>103595675
>None of the models are capable of doing high school level math.
are you braindead? even shitty gpt-4o models can do math nowadays
this isnt the base gpt era anymore

>>103595675
Lol, I bite (you are still retarded), give me the question that's so impossible.

>>103595651
>This one is insane
The word you're looking for is "scam". AI cant do math. Its not possible the AI has suddenly progressed so far from not being able to do math at all, like is the state of the art for all current models, to being able to do some math.

>>103595177
>intelligent board
no such thing
>AGI being realized
it's impressive, but still not AGI

>>103595651
>frontier math
is this the one that humans cant solve

>>103595708
math experts can solve them but it takes them a long time. and yes, obviously the average guy cant even think of solving them

>>103595683
Its not bait. I have tested it with a lot of basic high school level math questions and it kept giving retarded nonsensical word salad answers. Even for questions whose answers are well documented in many math textbooks.
>>103595688
You are actually delusional. why do you make such obviously false and retarded claims before testing them first? I have.
>>103595690
One example is asking it to show that R(3,4)<=9. Simple basic high school level graph theory question with well documented answers in textbooks, but none of the ai models can do it.
There are more questions.

doesn't really matter there aren't any jobs anyway

File: whisk_storyboard973c1745e(...).png (1.67 MB, 1408x768)

1.67 MB PNG

There is always this one schizos that claims ai can't do highschool shit but always ignores any requests to post a problem. Legit was causes this behavior?

>>103595693
You're pathetically retarded

>>103595708
>>103595718
https://epoch.ai/frontiermath/the-benchmark
Forgot the link. It has some examples but the actual benchmark questions are hidden to stop chinks from overfitting their 2b models on the questions and get a 100%

>>103595727
Holy fuck you're retarded. "Uhm I don't like the EXPLANATION huehuehue so it'll always be bad" are you really that retarded to think that people fall for your retarded bait? Give us a question with a clear solution you subhuman lmao.

>>103595177
1. This isnt agi or anything near it.
2. Your "intelligent board" has been infested with schizos for over 4 years
3. Nearly all real intelligent experts in the field are cautious and warning against agi in the first place

>>103595727
>R(3,4)<=9
Not HS math but just asked 4o and got the right answer
Now what nigger

>>103595740
Show me a model that I can use that can solve basic high school level math and ill change my mind.
You can't, because there arent any. You are being scammed.
>Le benchmark results!!!
Cool, where can I see what the answers AI models generated look like. Oh, that's right, theyre not available. As I said, youre being scammed.

>>103595727
>Ai can solve so much that he has to resort to "uhm I don't like this way, it isn't my way"
>>103595764
>Uhm ai experts warn against ai guys!!!
How are these utter retards allowed to post here? Afraid of ai and delusional about ai.

>>103595762
I don't get what the problem is. What's wrong with the question I gave? Or are you asking me for the solution because you can't solve it?
Here's another question: define the signature homomorphism and prove that it takes the value -1 on transpositions.

If it's so great, have it solve the latest advent of code question.
Protip: you won't and can't.

>>103595779
i have o1 api
ask HIGHSCHOOL math questions and I'll prompt it

>>103595783
Ah... youre one of the schizos. Got it.

>>103595778
>4o and got the right answer
Thats only because you're probably too retarded to tell what correct reasoning looks like. Go on, post the answer you got. I promise it's going to be wrong and retarded like it always is with these models, since they cannot think and dont understand anything.

>>103595779
Lol, I remember you, when people post what ai all sloves you always say "now try it won't work" despite people writing it works. How's life going? How close to suicide?

>>103593650
>o3

>>103595793
I asked a high school level math question.
Here's another one: suppose you have 100 boxes arranged in a row labelled 1 through 100. Prove that after swapping two boxes at a time, an odd number of times, you will not have returned to the original position. Do not use any results from basic group theory about the signature homomorphism, as this would make the solution circular.

>>103595788
The model it not out retard

>>103595806
>always say "now try it won't work" despite people writing it works
I dont remember this happening.

>>103595829
NTA but we didnt have anything like that in HS?
are you American or

>>103594668
>This thing would only be able to program concepts that it has seen
And the combination of every concept.
For example if it knows what "blue" is and what "brick" and "house" is, it can also know what a "blue brick house" is even if it never learned about blue brick houses before.
Now if you know the entirety of current humanity's knowledge, and it can combine all those separate concepts together, it's sure to program new concepts too.

>>103595841
So you're saying it's vaporware then? Weird self-own I guess. LLM cucks are delusional.

>>103595788
>you won't and can't.
it can tho?

>>103595165
These benchmarks are clearly aimed at potential investors. LLM companies need a continuous inflow of capital to keep the lights on. None of them are anywhere near profitable nor do they have good plans how to be profitable. They already set the price floor next to zero. The story they tell to investors is that they can replace expensive human labor with their LLM product because they know that's what opens C level pocketbooks. What developer hasn't seen a senior manager come back from a convention with some expensive piece of productivity software that a random VP drops on them, telling them the company spent a lot of money (really, the vendor took the VP out to a strip club and paid for a prostitute after) so they'd better make it work and be more productive with in.
Microsoft is paying God knows how much to restart the nuclear reactor on Three Mile Island (worst nuclear accident in US history happened there) so they can power more LLMs. How much does it cost to reactivate a nuclear reactor, especially one that is the public symbol for danger? Bet the regulators one day found a new story added to their home in appreciation for helping MSFT get away with that one.
LLMs in some cases can help with grunt work. In many cases, they create more cleanup work than the work they replaced. Each new model claims to have fixed this and it's never been true. But sure, this time will be different. Invest now!

>>103595850
Its a problem thats understandable and approachable by every high schooler with well documented short solutions that AI cannot solve

>>103595872
so it's not highschool math retard

>>103595868
That’s an excellent analysis anon, and I completely agree with your sentiment. I also believe that LLMs are largely a scam and are only useful in a few specific areas

File: Multioni.png (3 KB, 225x306)

3 KB PNG

>Coding is getting replaced before art, music, even trades
Is this karma, or is God trolling?

>>103593760
This will be the true reckoning, the chief complaint EVERYONE has about call centres (even more than queue times) is that the pajeets can't understand english and people can't understand their pitiful attempts at communication.
Total jeet market genocide is coming.

>>103595918
Art has already been replaced, the sheer volume of AI art I see on the web dwarfs the amount of actual human art. Music is well on the way since Suno/Udio are actually very good and only really need "fidelity" upgrades to instrument quality.
Trades are safer than people think, I had my old kitchen redone this year and no AI model could ever solve the problems in that shitty little room that the workmen even struggled with. AI won't be able to do shit when it accidentally hits an electrical plate or wire in the wall put there in the 50's.

>>103595918
It's moravec's paradox. Just happening 10x faster than anyone thought it would

>>103595634
Five years ago that list would've been ten times longer

>>103595634
Yes, and that's how the elites will keep the masses away from revolution.

Some UBI to guarantee food, some VR slop to keep them busy, and the occasional crypto scam / social media popularity contest that makes them think they could also join the upperclass with enough effort

>>103595974
It really wasn't. Artists are still entering the space and getting viral reactions. Gaming companies avoid AI art like the plague, that volume of genned images continues to fail to get the same engagement as human art.
Same with music, people simply ignore AI generated music, the space continues to produce and promote human musicians.
>Trades are safer than people think, I had my old kitchen redone this year and no AI model could ever solve the problems in that shitty little room that the workmen even struggled with. AI won't be able to do shit when it accidentally hits an electrical plate or wire in the wall put there in the 50's
That's the point, AI seems to be most capable with working with characters, not graphics, audio, let alone interfacing with the real world.

>>103595177
Its just /g/ has lot of reddit trannies seething about jobs and communist nonsense

>>103596045
wjat comes next is communism retard

>>103596056
Starvation from lack of foods in the stores? LMAO Commies want to take credit for the creations of capitalist society.

>>103593650
Shit, all the jobs in competition coding might be in danger

>>103596045
Communism is literally the only path forward (apart from techno-feudalism with 99.9% of people enslaved)

>>103596103
no farming will be automated like everything else. The state will control everything. COMMUNISM.

>>103596115
>state will control everything
Thats the commie dream ofcourse. America rejected that and elected Trump. AIs under private individuals and companies are the way forward.

>>103595531
You're stuck in 2018 dude. Robotics is advancing at an even faster rate than LLMs, we will have lightweight and dexterous humanoids before 2030, hell even before 2027 at this rate

>>103593650
learn to uhhhhh

>>103596130
if nobody works there is no private anything . the ones that own the robots and ai become the state.

>>103596146
Supply demand

>>103594668
that's the whole point, o3 is capable of extrapolating new solutions from previous data
it always used to be able to do this, but would always give terrible answers
the new model is much better at this than before
with the rate of progress, it's only a matter of time before AI can solve unsolved problems
the implications of this are huge, absolutely beyond belief
attach efficient forms of these AI models to mechanical forms and we unironically have a robot problem

>>103596156
supply and ration. the demand is determined by the ones that own things. if nobody works they dont need your demand.

>>103596157
remwmber when you retards claimed that encryption was useless now because of muh agi?

>>103595177
is sci a fun board? i'm sick of the blatant psyops and sequential retardation on the website

>>103596165
what about quantum computers?

>>103596164
>ration
Communism was defeated with the election of Donald Trump as the 47th president of the United States.

Maybe you didn't hear the news.

File: download (28).jpg (8 KB, 275x183)

8 KB JPG

>>103596186
communism cant be defeated just like capitalism couldnt be defeated by the luddites

>>103596202
Purging the WEF is necessary step. Relocating DC workers to other states like Arizona/Nebraska/Florida is a start. The swamp must be drained and not allowed to manifest.

>>103596211
the only solution i can think of is that the dumb working class must create an equivalent or superior open source ai along with the tech and the chances of that are slim or if you want to go after them now itd be the time before you have to fight terminators. idk.

>>103593650
OpenAI suicided the whistleblower who was talking about how they use everything and anything for training. They're just training on the solutions now to keep up the endless money pit of a dead end technology.

>>103596233
>open sores
Worker class rejected tranny communism. And elected Trump.

>>103596252
Have you considered that he was simply depressed?

>>103596268
Lots of depressed whistleblowers in corporate America, don't worry Sam your fellow oligarchs are all doing the same.

>>103596268
Depressed with bullet to the back of his head

File: Screenshot 2024-12-21 140412.png (674 KB, 609x614)

674 KB PNG

>>103596285

>>103596290
You know police investigations exist to determine these sorts of things? If it's not a suicide the police are able to determine it. You're just a delusional conspiracy theorist.

>>103596177
>what about quantum computers?
what the fuck about them?

>>103596300
You realize the democrats are protecting the Epstein client list right? They've refused to release the list. Same 20+ J6 FBI insiders who were in the crowd. And how the investigation commission hid all the evidence of their wrong doings and created the false "insurrection" narrative right?

No one trusts anyone unless they've fully embraced the castration ideology and have castrated themselves

>>103596295
>>103596300
Answer the following question: Did Epstein kill himself ?

>>103596167
It's better than the others but it also has been going downhill in recent years

Personally I can only stand /sci/, /his/ and /ck/ these days

>>103596167
Sci is schizo central, an offshoot of /x/. It's where schizos come to post about how they discovered a solution to the universe or whatever other word salad happens to bless their heart that day.
It's also being invaded by /pol/ 24/7, with constant anti-science, anti-math threads. The mods allow and encourage it.
All the people remotely competent at science and math have left the board a long time ago.
It's a shithole but if you're a retard without a brain and with a lot of time to waste I'm sure you'll fit right in there.

>>103596316
Based on what I've seen looks like there was foul play involved.
Nothing to suggest foul play in the openAI case.

>>103596382
this is very irritating. sci is the anti science board.

o3 looks like the happy merchant

>>103594144
based, this is how it should be by the way

>>103594844
idiot kek

Why are people retarded enough to believe OpenAI isn't blatantly lying?
They have been hinting at having an "AGI" since last year. The evidence was just some made up shit then and it's just some more made up shit now.
We have been in a panic about AI taking art jobs, writing jobs, and coding jobs for a year now and a ton of examples of that supposedly happening have also been outed as fake (and gay).
o3's commercial release will be just another error riddled joke. Remember OpenAI's promise to make Sora more uncensored than previous models?
You probably forgot already.

>>103596457
Why would someone lie on the internet?

>>103596441
You will be in the underclass unless you're a jew btw

File: 1719339867693221.png (634 KB, 585x888)

634 KB PNG

>outsourcing to india will replace software developers
>WYSIWYG will replace software developers
>low code will replace software developers
>LLM's will replace software developers
Anyone making these claims has clearly never worked as a software developer in any serious capacity. Protip: software development isn't about writing code.

how tf aren't we seeing uprisings or cyber attacks against these companies?

>>103596540
>Heard about the guy who fell off a skyscraper? On his way down past each floor, he kept saying to reassure himself: So far so good... so far so good... so far so good.

for something that costs 50 thousand times more, that isn't really all that much of an improvement is it

My job is mostly spending hours in a debugger to figure out what's going on in a 100k+ line codebase and then remove/modify a single line to fix the bug. Should I use AI to generate this single line to boost my productivity?

>>103596540
only person ITT who has ever worked as a programmer

>>103596926
Do you think whatever o3 is, that it's optimized to the metal? Bringing down the cost by 10000x will be a walk in the park

Human is far cheaper than o3

>>103597060
Why didn't they then walk in the park before shipping?

>>103595177
ChatGPT is bullshit

https://link.springer.com/article/10.1007/s10676-024-09775-5

>>103597229
Maybe because

1) they haven't shipped it yet
2) it's a fucking time race against other 3 companies and they must show progress as soon as it's made to retain investors

>>103597271
I am not reading that. I am here trying to help a bunch of retards understand where the world is going, at unprecedented speed

If you're too dumb to listen to me, don't say you werent warned

>>103596022
>meanwhile, me who can't stand modern music and my favorite song "Deep Night" being AI generated
Or you just want to believe so because it's easier to think this technology doesn't and won't have any impact so you don't have to adapt and change. I remember when people used to say the same about computers, telling us we were wasting our time learning how to use them, that it was just a "toy", that we were "childish" to think these would be anything but another fad. History repeats itself it truly seems, humans can't stop being morons who make the same mistakes over and over again.

File: 1723284537800.png (202 KB, 718x589)

202 KB PNG

>>103597296
>claims to be a /sci/entist
>can't even read

>>103597317
False equivalence
The scale of investment required to train an AI model dwarfs anything related to computing R&D and manufacturing.
These companies behave like sharks running purely on investor funds and data, the moment one or the other runs out, they die.

>>103597296
>where the world is going, at unprecedented speed
Nowhere

>>103597429
No, through the technological singularity. Listen I also used to think Kurzweil was a massive faggot when I first read about him 12 years ago, but turns out he was right, even conservative.

His estimate for 2029 will happen by 2026

>>103594742
codemonkeys are delusional if they think designing software is a more cognitively challenging task than solving frontier math problems
I'm happy you are going down tbqh

>>103597620
Current gen AI can do neither.

>>103594755
>i'll program because i want to! because it's fun
programming will be the hobby of autistic shut-ins again