>>110257869
imaging thinking people will pay for Google search. those Jeets are showing us a new level of stupidity

the funny thing is the red square will be the only people who have jobs by next year thanks to AI

>>110257891
Claude, lift this box

>>110257869
The only thing I have ever paid for on the internet is goods. Never services. Not going to start now.

>>110257869
this is profitable because companies and political parties pay billions to the AI companies to train them with biases so they give you fake information and mislead you and control you
they're planning to make newspapers and wikipedia obsolete, you only ask the AI

>>110257869
I would never pay for AI, because i am not a fucking retard.

You offer it for free, or fuck off.

>>110257869
Yeah there's billions of shitskin hut dwellers in the world who have never seen a computer or a functioning toilet. What's your point?

>>110257987
actually I asked, they only mislead (You) specifically

File: 1699602938939607.jpg (144 KB, 1280x720)

144 KB JPG

>>110257998
Came here to post this.
I don't need to rent extra brains. I'm happy with mine.

Most people are retarded and faggy, Is this sustainable?

For any ai shills here, what are some actual benefits of paying for AI? Do people really vibe code shit and sell it? Do you generate money from AI?

>>110257869
The Chinese offer it for free, so you have to offer it for free as well.
You do want to win the SI race, right?
China will win if you don't do it for free.

Every SI company that is uncompetitive against China is a national security threat and needs to be dissolved and given into the hands of a patriot.

>>110258053
>Do people really vibe code shit and sell it?
yes but i would say there are different levels of vibecoding. i have products already designed and was working through them before AI could write code. i'm not saying "make a tetris clone" to the prompt and coming back after 4 hours to see what is done. i give it the design docs and "discuss" the docs and specs and get it to suggest improvements and issues before it begins any code creation. it is actually quite tedious to oversee it step by step but it is far quicker than writing code myself.
i don't start my proper work until this afternoon because i have my own schedule but this morning since 10am (it is 12pm now) i have had prompted the equivalent of a couple of weeks of evening and weekends work, all while i have been reading 4chan and other internet.
it is relatively niche audio software, previous programs sell 1000 or so copies over long term.
i could do it all myself as i have the knowledge and experience, but i couldn't be preparing the products i am preparing now all at the same time over such a short timescale.

>>110257869
https://thesouthfirst.com/business-tech/india-is-chatgpts-biggest-market-globally-user-base-outnumbers-us-sam-altman/

>>110257869
They will paint it as growth opportunity, regardless of the fact that 95% of the usage will be equivalent to a slightly improved google, until they enshitify it.

>>110258053
The only people generating money with SI are the ones who resale it. Like platforms for ERPing with AI partners or image generation.
Those are also the ones you see on twitter advertising AI and how rich it made them.

Wageslaves, who optimize their prompts to waste the max amount of tokens to meet the corporate target, aren't any more productive.
Ask proton about it.

Then you also get the AI psychosis victims, who "ship faster", but whatever new feature they ship is unfinished and broken and they live in an infinite feedback loop of continuious development where they never reach any target.
Those people are arguably less productive.
All the personal examples i know, produced more valuable things before they fell for AI. But they don't want to accept it, because "it is faster".
I rather drive 40 mph straight towards the target, then drifting with 200 mph in a circle, constantly missing the goal.

I don't pay for dice rolls.
If they could somehow guarantee consistency then I'd consider it, but knowing that you put $20 down and might get no value out of it, makes it not justifiable for me.

>>110257869
i cant imagine anything more cucked than paying for an AI chatbot

>>110257869
This explains why I'm making so much money and the only thing I've changed is incorporating AI into my workflow.

>>110257891
>all jobs involve coding
bot post or human with bot-like intelligence?

>don't pay
The free chatbots cost pennies to run and those companies are harvesting and sorting fine grained data that google couldn't even dream of 5 years ago. You have no idea how many people use chatbots as their fucking best friend already.

>>110259838
So why isn't AI profitable

>>110259858
The subscriptions. While it may be lower now, for quite a while your $100 sub gave you around $8000 in usage for Claude if you went by api pricing.

The problem is the moment they get rid of subscriptions is the actual moment AI collapses. They’re trying to drag it on to hopefully get AGI or whatever the fuck before they have to kill subscriptions to stay afloat.

>>110257869
>paying to use some other persons computer when you can host a local model
cucks

my boss pays for codex, I would never get a subscription to this shit at home

>>110257869
>Each dot
>Only dots are decimals, ends of sentences, lowercase i, monospace zeros, and some weird separators
what did they mean by this?

>>110257869
Why even bother including the entire planet? There's at least 2 billion there that quite literally do not even matter to any tech discussion.

>>110260540
It's literally how all this scam is pitched. Projections are made on assumptions every person, alive and dead, will run 1000 agents at all times. You can't get to 6 gorillion percent increase in GDP otherwise.

File: 1779604701950719.png (427 KB, 1694x1330)

427 KB PNG

>>110257869
nope, what the most sensible of not-bots & not-shills having been telling you for the past 4 years at least

its a tool, its not general intelligence, the trend is smaller and local, services will be useful but no one is going to pay the big money these dumbass scammers are expecting

>>110257891
>the red square
they're buying dollars, each for a cent
anyone smart will abuse that until it all crumbles

>>110257869
Where is the local dot?

>>110257869
No, it absolutely isn't.
Majority of AI use isn't WORTH paying for. It's just a replacement for a search engine, formatting tool, etc. The tech is definitely useful, and definitely here to stay, but the financials aren't there.

Eggheads in silicon valley made a really cool piece of software, and are scrambling to try and find a way to market it to people who don't see a need for it. All while making their lives worse by shoving it in their face causing backlash.
It's tragic if it wasn't so predictable.

I pay 20$ and steal 1000$ of worth it from Anthropic and I will immediately leave for open source chinese models is it to become more expensive than 30$ a month at any given time.

>>110260965
>definitely here to stay
Only for braindead usage like search, formatting, and unreliable autocomplete. For any actual use it's too expensive to run, and not worth paying for.

>>110260988
No way in hell it's only 1000$.

>>110257869
that is a staggeringly high adoption rate for a technology that has been around for a couple years

>>110260990
>search
I always see that argument but Google has worked better 20 GODDAMN YEARS AGO.

>>110261009
>that is a staggeringly high adoption rate for a technology that has been around for a couple years
Is it though? For something you can access from a computer/phone and buy immediately with your credit card?
Televisions and smartphones had to be bought and were very expensive in their early adoption years.
AI is dirt cheap for the customer and most people don't give a shit about paying for it... so where it's the high adoption rate?
Compared to previous technology, adjusted for how easily accessible and "cheap" it is, it's underperforming and clearly unpopular by all metrics.

ye, aa-absolutely, its like with content hosting. do you pay for youtube? for github? for file store? most of those are maintained by US companies.

> current global data creation is estimated at quintillions of bytes per day

> from 2 zettabytes in 2010 to 181 zettabytes in 2025

>>110260988
It's is the average "AI power user". As seen by all the usage limit complaints.

>>110261009
That's what happens when you put it front and center everywhere, for free. You make it run by default on a search engine against the users will.
But just usage doesn't pay for a, in computing terms, exponentially more expensive service.
>>110261050
Yeah "profitable adoption rate" is miniscule.
>>110261011
To be fair perplexity was pretty good. Which I used to access expensive models for free with a free pro plan from revolut.
But yes google search declined very noticeably from 2018-2022 ish, youtube search too, unable to find stuff because it shows you what THEY want you to see not matching results. And removing reverse image search.

It depends, so far the 20$ plan has been worth it because it has managed to last me for days on end for any type of work.

Beats a Netflix subscription. And I don't pay any other monthly bill. Otherwise if they raise either the token price per dollar or the sub it self? I'm out and back to free. No one is paying 100$+ a month for that.

Or I might save up for Local AI.

>>110261159
I've been running a AI service since before chatGPT. It's been clear the whole time DIY can't even compete with API usage, meaning it's subsidized because the OSS models are cheaper per unit of performance over API too.

Compare it to cloud storage or cloud compute where going DIY has paid for itself in a year or so, the whole time.

> if they raise either the token price per dollar or the sub it self?
The future is a $20-100 ish sub, just to capture free money from the inactive users, and then strict limits so you have to pay a lot more ($50-200/day) for heavy usage.

>>110261069
Video compression aside, none of that comes even close to bespoke hardware and power requirements

>>110258053
I think the advantage is getting access to the smarter models, the free tier is always several levels dumber than whatever flagship they have

>>110261302
This is eternally the business case of openAI/anthropic.
level 0.95 (out of 1) AI they will give away for free.
For those 95% open models are cheaper so at scale, they use those.
They are hoping that one "frontier user" can way overpay, pay for 20 free users and their compute, the whole company and their salaries, model training...

Extremely risky business model.

local is the answer

>>110261447
local is antisemitic

>>110257899
that would be anti-semitic and would violate my guardrails

>>110258031
>>110258023

>>110258137
deepseek isn't free for coding, really

File: rfxlhjb7ple51.jpg (96 KB, 1080x1080)

96 KB JPG

>>110257869
>Most people don't use AI,
Most people didn't drive cars. That changed fast.

>>110257869
The jeet summarizes it quite nicely:
"The products aren't good enough to sell people on the costs. You cannot tell people that you have to listen to the whine of a data center in their backyard for AI overviews and some of the most hilarious deep fake slop in history. It's not worth it. It just isn't worth it to anyone."
https://www.youtube.com/watch?v=-U4thshDTL8&t=422s

>>110263978
>The products aren't good enough to sell people on the costs.
I've used AI to replace several artists and save $1000s of dollars.
I get way more money's worth from prompting my own pictures, music, videos instead of paying some mentally ill troon to draw stuff.

>>110263957
everyone would have immediately used cars if they were free

>>110263996
The people refusing to use AI will simply be culled instead.

They'll lose their jobs, they'll lose their mortgages, they'll be homeless.

>>110257869
1/6 of the world is a lot thoughbeit

>>110258053
I wouldn't say I'm an AI shill, but I do use it quite a bit. I'm an Analytics Engineer at a megacorp whose idea of AI is copilot, so I vibe-coded a custom harness that I've been modifying to automate 99% of my software tasks so I have more time with my family
I also have a leftover Claude subscription my wife is using for her Master's program that is quickly becoming useless, and I buy ~$20 deepseek credits monthly for projects. I'm looking to replace both with locally hosted models when the next hardware refresh cycle takes place
Tdlr:
>College Research, Work, Hobbies

Is it worth it to become a red square?

>>110263957
>hurr AI is like a car vs horse
You keep spamming it, so it's wrong on principle. Good day.

File: lol.png (732 KB, 1271x1476)

732 KB PNG

>>110257869
ftfy

>>110257877
Yet Google still makes money. Just wait until they figure out how to inject ads into it without affecting the prompt…

>>110264106
>>110264106
Every non-AI person is 100x slower than any AI person.
In 5 years the gap will have widen making employment (for non-AI people) impossible.

>>110257869
More people use AI than that if you count Google's search result summary.

>>110264308
they why have AI persons achieved jack shit?

File: 1asas.png (921 KB, 881x1491)

921 KB PNG

>>110264363
The math community got mogged 2 months ago.

>>110264368
that's just brute-forcing shit with millions of dollars of compute

>>110264356
I wouldn't really count that though. It is "AI" but a severely lobotomised version of it that's trimmed down and worse than Gemini Flash even.

File: 9568956895.png (115 KB, 1671x968)

115 KB PNG

>>110264381
The newer models can solve math problems faster.

Hence why OpenAI released that big batch of 700+ math solutions.

>>110264389
openai has withdrawn every math proof that an actual mathematician looked over properly

I'm sorry but you fell for a very expensive PR campaign

File: 1983958495454.png (217 KB, 1458x1281)

217 KB PNG

>>110264397
Wrong, asshole.
https://github.com/openai/math
Go put the fries in the bag.

>>110264397
It's worse than that because the mathematics community actually understands the potential for AI to solve maths problems and wants to work with them and demands the transparency that the community expects but they won't do that.

They won't share the inputs and prompts.

File: 1789941882073994.png (666 KB, 874x1328)

666 KB PNG

>>110264410
What transparency?

When OpenAI solved Navier-stokes, their first reaction was to chimpout and call for a ban on technology.
They're spineless pussies and should be treated as such.

>>110264404
these are full of very obscure problems that very few people have decent domain knowledge of and very bad hard to follow proofs
no expert really had the chance to properly look at all the slop openai has spammed yet

File: schizo math.jpg (3.66 MB, 2121x1184)

3.66 MB JPG

>>110264424
>no expert really had the chance to properly look at all the slop openai has spammed yet
The "experts" are the same as the artists who said AI would never draw hands and then ran into a cave.

AI is smarter than most humans now. You have absolutely zero leverage to stop this thing.

>>110264415
Their reaction is justified to be honest.

Imagine your an expert in the field of mathematics trying to understand all this and Open ai just dumps a very badly formulated PDF on your desk that almost certainly doesn't contain all of the information OpenAI has internally.

Those shitty "proofs" aren't worth jack shit because what you actually want to understand is the workings out the AI is doing and how it came to its conclusions but of course all of that is proprietary information they keep secret and won't show anyone.

>>110264437
non sequitur

File: caveman-starting-a-fire-x.jpg (201 KB, 1200x800)

201 KB JPG

>>110264438
If we followed your logic, Cavemen would have banned the use of fire because they didn't understand it.

You're only trying to stop progress and science because you have a fragile ego. That's not my problem. Go find another job.

>>110257869
most people don't use airplanes
most people don't use nukes
most people don't travel to the moon
most people don't dive deep into the abyssal ocean
most people don't build a skyscrapper

>>110264457
Cavemen didn't hoard the ability to make fire for themselves. They shared it with others. They didn't document it in cave paintings but then never tell anyone how the magic works.

>>110264475
>They didn't document it in cave paintings but then never tell anyone how the magic works.
Anyone can use an AI model and work on math problems for themselves.

In fact, this only makes OpenAI's argument even stronger.
In 2 years from now, any AI model can solve a millenium math problem and you can read/ask it yourself.
If you still bitch/whine, that's on you.

>>110264126
>Yet Google still makes money
Because Google search doesn't require the assloads of compute LLMs do, retard.

>>110264028
the next hardware cycle... uhhh yea everything is fine same as it ever was

>>110264488
>Anyone can use an AI model and work on math problems for themselves.
Few people have the millions of dollars to burn on compute like OpenAI did.
This wasn't just some 30 minute ChatGPT sessions, mate.

File: 1_YrTcGwOdfkKNbmFtC1ar4Qa.jpg (255 KB, 1026x1283)

255 KB JPG

>>110264498
>Few people have the millions of dollars to burn on compute like OpenAI did.
You idiot, AI models have gotten both faster and inexpensive over time.
It is actually the first technology to make such exponential improvement for the masses.

Hence why I said "try it in 2 years".
But by the time you prompt the math problems, AI will have solved bigger things like curing cancer.

>>110264510
If AI solves cancer the first thing scam Altman will do is sell the drug to a private healthcare company and it'll be as unobtainable as RAM.

>>110264510
>crystal ball shit
Yeah, that's what cavemen said when they discovered fire.
Sure it takes the largest tribe in the world to make a tiny flame but just wait a few years

File: 953.gif (198 KB, 500x276)

198 KB GIF

>>110264517
Good. Then it means you will be stuck with AIDS while AI Chads go live forever on the moon.

>>110264523
That's the idea. There won't be any human betterment. I bet he wouldn't even release any of his discoveries in an open patent pool.

File: 010101212.png (105 KB, 1080x1687)

105 KB PNG

>>110264519
It's well documented on benchmarks.
AI models are cheaper to use while being more powerful than last year.

>>110264539
Benchmaxxing is the opposite of useful practical intelligence

>>110264539
We don't even know the true costs that OpenAI has.
None of that is open. OpenAI shares jack shit.
So no, nothing is "well documented"

File: goalposts.jpg (307 KB, 1056x594)

307 KB JPG

>>110264625
Keep moving goalposts jackass.

AI models getting cheaper is consistent with how open source models have also lowered their bills.

>>110264634
Nobody cares about your shitty benchmark.

Only an idiot would pay for SI. The SaaS companies can choose: Give it away for free, or die.
If they don't adapt to free models, they will get left behind.
OpenAI and Anthropic are outdated.

>>110264437
left
>chill with friends
>organic sunlight for the majority of your day
>smoke weed and say dumb shit
>can seclude yourself for privacy
>jaw dropping natural scenery all around you
>eat home-grown, real food with recipes passed down from generation to generation
>cool clothes
right
>study slop
>bright fluorescent and LED light in your eyes 24/7
>only entertainment available is government approved Netflix© and Sweet Baby Verified Stamp of Approval ™ video games
>everyone forced into glass buildings, curtains outlawed
>Flock Safety ® drones patrol the streets and every home
>look outside your window: huge hologramic ads for the latest slop
>can only order food, everything other than UberEats Instant-Delivery® deep-fried M&Ms is sold out
>same t-shirts and lab gowns for everyone

>>110264634
how is pointing out that we can't even know shit is getting cheaper like you keep claiming moving the goalpost?

>>110257869
It doesn't need to be sustainable. AI roll-out today is about making it key to infrastructure. If the average mudhut dweller uses it is completly irrelevant to its actual purpose. Anthropic and OpenAI are effectively shell companies that only exist to build the infrastructure and launder the money so the US goverment can pretend its the operations of private companies and not them.

>>110264688
>Only an idiot would pay for SI.
As much as I enjoy open source models, they are still behind frontier models by at least 6 to 12 months.

It's why Claude was able to do all those cool tricks this year like modding video games. None of the "free" ones can match that performance yet.

And even by the time they do, AI from OpenAI or Anthropic could literally create entire games anyway...

File: ezgif-2209cc84445c0192.jpg (227 KB, 1200x720)

227 KB JPG

>>110264692
You forgot that hippies are always staring down the barrel of a gun. I.e, they are the weak and powerless ones.

If the robot soldiers start shooting, your organic math won't save you.

>>110264790
>behind frontier models by at least 6 to 12 months
Yeah, who can wait 6 to 12 months??

>>110264790
Frontier models are behind Open Source models by an infinite factor.
One is free, the other one wants money for making a dice roll. It's a division by 0 error.

LLM-as-a-service is simply not a sustainable business.
>b-but maybe i can produce an output in one prompt that an open source model would produce in two... and the open source model will be able to do in one as well in a few months
What argument even is this, lol?
What did vibe slop companies achieve in the past months? How did Windows improve? How did Proton improve?

Also the growth is logarithmic, each newly released model has a smaller gap to the previous one. The gap keeps getting smaller and smaller.

OpenAI and Anthropic need to offer their LLM for free or get left behind.
They will be in the trash bin of history (and with them the whole US economy), while everybody moved on to the better competitor.

>>110264835
>OpenAI and Anthropic need to offer their LLM for free or get left behind.
why offer a model for free when you can just buy all the GPU's and shove them in a warehouse, indefinitely
if you deny the people the ability to run a local model they have to buy it from you

>>110264790
I sell cars, and my competitor gives out free cars.

But my competitor is 6 months behind!
Do you really want to wait 6 months to get this specific model for free... or get it right now while paying for it?
Also when you pay me, i have a gps tracker in your car and know where you go at what time.

>>110264842
In that scenario, you get local businesses buying the hardware and selling it for cheaper.
Like if it takes 5 million USD to get hardware to run Kimi or Deepseek (or whatever else is the top free model)... it is still profitable for some mid-sized business to get that, and they can then rent that out for cheap.

Anthropic and OpenAI still die in that scenario. It is simply not competitive.
They have to offer their product for free or get left behind.

File: 1774162249096911.jpg (131 KB, 1088x960)

131 KB JPG

>>110264790
Mythos preview was out in the Feb and none of the open source models released this year can match it. If what anthropic said about blocking distillation is true then this is as good as open-models will ever get. They don't have enough capital to do post training on their own.

>>110257869
That's precisely why the bubble hasn't popped yet. "Look at all this untapped market, imagine if even 10% paid for AI. Given that only ~5% use AI at all, and only 1% of that 5% pay for it, if 100% use AI with 1% paid that would be 20x current revenue!"

>>110264883
That's what you think, because you didn't try them.
You fucking Luddit.

>>110264874
>Anthropic and OpenAI still die in that scenario. It is simply not competitive.
not in the long term but it is the strategy that is being deployed.
https://www.businessinsider.com/amazon-nvidia-ai-gpu-data-centers-power-shortages-2026-10
Amazing is buying billions of dollars worth of hardware and sticking it in a warehouse. they don't have the electricity to power the chips and probably won't by the time those chips are out of date.
and i assume the same strategy is being deployed by all the different companies. it is the act of a mad man and yet here we are and it is all being funded by venture capital, pure insanity.

>>110264883
>If what anthropic says is true
lmao

>>110264883
I recently had a problem that Claude produced 500 lines of code for, while Deepseek achieved the same result in 100.
Claude fucked around and created methods for shit that was already available.
It was astonishing. And i see this more and more, that Claude simply throws useless slop into the codebase, redundant code.

And Deepseek isn't even the best free model...

>>110264910
>Claude fucked around and created methods for shit that was already available
Wageslaves, who got told to use AI by their management and who got their performance measured by the amount of tokens they use, need exactly that.
They need convoluted large outputs. Completely irrelevant, whether or not that is needed.

File: 1793611512970897.jpg (235 KB, 1280x1264)

235 KB JPG

>>110257869
>Most people don't use AI
Most people have no use for AI, they can think for themselves. They can take away all the PC components from people they like, and they will still not need AI.
The, "everyone will need, and be using AI in their homes,", are the delusions of desperate men, who got high off their surrounding circle of sycophants.

>>110257869
i got frontier and no pay.

>>110264893
>If I can't use that GPU you can't either
It's so stupid and the reason GPUs cost an arm and a leg, more like a kidney, and no consumers can afford one.

I've been looking at building a PC for someone for £850 (that's the budget they've given me) and you can get fucking nothing for that. It's so depressing. I'm just going to give them a spare rig I built for myself but don't use. It's worth way more than that (9800X3D + my old 6700 XT and it has RAM and storage in it!) and they'll get something nice out of it.

>>110265214
can't you find stuff on the used market, like an old workstation and a gtx 1080 or a rx6600 from china or something?
buying new, expect your ssd, is for suckers

>>110265232
Nope, those cost a metric fuckton, too.

>>110265232
I could. An RX 6600 8GB is £200. That's absurd in my opinion. It's possible but the storage and RAM is a significant expense.

I don't use that PC (built it and left it down my brother's house for when I visit but I rarely do) so it'll basically be a very generous gift and I'll get £850 out of it.

>>110265248
>it'll basically be a very generous gift
your a good person anon, i am sure they will appreciate it
but that is also why i don't throw away any of my computers. people used to laugh, not so much anymore

>>110257869
Why would I pay for something that can't construct anything beyond a basic calculator without either shitting itself and starting to loop with constant errors in its own code or the chat window freezing because the retarded jeetshit implementation can't handle text beyond a certain total length.

File: bait.png (63 KB, 1024x1024)

63 KB PNG

>>110257891

>>110257869
>grey squares:
>elderly (will never use AI)
>children (will never use AI)
>poors (can't use AI)
outlook not so good