[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
/biz/ - Business & Finance

Name
Options
Comment
Verification
4chan Pass users can bypass this verification. [Learn More] [Login]
File
  • Please read the Rules and FAQ before posting.

08/21/20New boards added: /vrpg/, /vmg/, /vst/ and /vm/
05/04/17New trial board added: /bant/ - International/Random
10/04/16New board for 4chan Pass users: /vip/ - Very Important Posts
[Hide] [Show All]


[Advertise on 4chan]


File: images.jpg (7 KB, 200x200)
7 KB JPG
Bullish or Bearish? You decide
>>
poopish
>>
>>62936798
Zeroish
As in it's going to zero
>>
>>62936798
dumpish
>>
>multi-trillion dollar "creative roleplay" text generator
lol
lmao
>>
File: jvhn3tvi2s2i1.png (78 KB, 1226x364)
78 KB PNG
>>62936798
Riemann Hypothesis is now solved.
Anthropic won.
>>
>>62937345
FAKE NEWS
>>
>>62937345
and this makes money how?
>>
File: kike.png (377 KB, 578x684)
377 KB PNG
>>62936798
it's not looking good for ant baggies
>>
>>62937402
>Adam Smith writes a book calling for Capitalism.
>"But how does this make money?!?!?!!"
Seriously, you anons are a joke.
>>
>>62937345
good goy, believe everything you hear from a desperate company in the red
>>
>>62937345
Fakest larp ever award, AI psychosis award.
>>
File: 1795103718139320.png (269 KB, 2514x2298)
269 KB PNG
>>62937345
omg brb investing
>>
File: 1795178010770522.jpg (341 KB, 1080x1695)
341 KB JPG
what doing?
>>
>>62937318
HAHAHAHAHAHAH
>>
File: 1asas.png (921 KB, 881x1491)
921 KB PNG
>>62937483
Just like Navier Stokes right?
That's TWO millenium math problems solved in a fucking year.
Which by the way, the prizes for solving them are worth millions of dollars.
Now shut up.
>>
DUDE MATH LMAO
>>
File: 1696542847851639.jpg (157 KB, 768x1024)
157 KB JPG
>>62937521
>>62937345
>The poopoo hypothesis has been solved!
But how does this fix the fact that they are burning tens of billions of dollars a year and have no moat?
>>
>>62937553
That nigger can't accept that two things can be true at once:
1. AI is a revolutionary technology
2. It's not profitable nor a good investment
>>
File: _1Untitled2.png (630 KB, 891x968)
630 KB PNG
>>62937553
Every company is laying off employees by the thousands.
AI has made you completely obsolete and corporations do not need you anymore.
But keep sticking your head in the sand. They're going to be worth trillions in 2027 while you eat out of dumpsters.
>>
>>62937521
Wow a bunch of numbers lined up with a bunch of other numbers amazing. How does this make money again? Or does it just fellate 0.00001% of the population?
>>62937345
>Riemann hypothesis
Ok but how does it make money to justify the share price? They made the information public and can't patent anything of value.

P.S professors are mad because the AI can't show it's working. An understanding of how the math was done is extremely valuable so that it can be built upon
>>
>>62937483
This. I can tell by the megalomaniac style of writing alone that he's some easily manipulated mathematician that Anthropic uses for headlines when its numbers are tanking.
>but muh Lean proofs
Lean had a bug that allowed it to verify a false proof of the Collatz conjecture. As with the OpenAI Navier-Stokes case, for all anyone knows, these models are brute-force hammering the Lean engine to get bug-based verifications. Nobody has verified the Navier-Stokes proof, let alone (lmao) 250,000 pages of Lean.
>>
>WE SOLVED THE RASENGAN HYPOTHESIS
>stock goes even lower

UH OH STINKY
>>
>>62937521
Hasn't been verified, midwit.
>>
File: ezgif-543f12df6d45e3e3.jpg (213 KB, 1920x1230)
213 KB JPG
>>62937588
The biggest cost to every business on Earth is human payroll.
Anthropic is leading the way by offloading those costs to AI that can work forever and never need breaks.

Every company on Earth will want to buy this model while firing useless employees.
And their moat? They will achieve RSI that makes this happen faster. Nobody can match that strength.
>>
File: 10956045065.jpg (3.66 MB, 2121x1184)
3.66 MB JPG
>>62937632
AI is solving new math problems every week.
Humans will never be able to catch up and beat them at this.
The job of human mathematician is officially over. AI will always write better math from now on.
>>
>>62937640
So basically you're saying that this stock succeeds if we get the compute-equivalent of perpetual motion. Have fun with that, retard.
>>
>>62937686
Exploiting Lean bugs isn't solving math problems, midwit.
https://x.com/gro_tsen/status/2082483878480977959

>So, someone came up with an AI-generated formal proof, in Lean, of a solution to the Collatz problem, and it turned out that the “proof” was merely exploiting a bug in the Lean kernel (allowing you to prove anything).

Show us where that Anthropic red hurt your wallet... kek
>>
>>62937696
>t the compute-equivalent of perpetual motion.
No one can stop AI.
Not Russia. Not China. Not basement cucks.
So yes, it is basically perpetual motion.
It's the closest thing to real life cheat codes.
>>
>>62937686
>create proof generator that only a handful of mathematicians care to pay for
>proof generator makes mathematicians obsolete
bravo dario
>>
>>62937686
To be fair there's no niggers on the hippie side.
>>
File: IMG_3861.jpg (82 KB, 675x680)
82 KB JPG
Risky betting on a single AI winner. Logic says Google, but Anthropic or OpenAI could come out on top and make Google into IBM of software.

But also, the Chinese are trying to make tokens cost as little as possible. American frontier labs would need to pull a few rabbits out of their hat when it comes to pharmaceutical research or something similar.

>t. someone who pays $100 a month for Claude
>>
>>62937751
They still have aids from all the homo gay sex though.
>>
>>62937447
One of the most non sequitur things i have read in a while.
>>
>>62937754
Chinese models are a year behind American ones. And the new anti-piracy measures from Anthropic make it impossible for China to distill their more powerful models.

China is cooked.
>>
>>62936798
will have lost 95% of ipo price by end of 2027
>>
>>62937515
Dario and the EA sex cult are laughing now. ANTHROP\C's cash shot up to nearly 200bil yesterday and now retail scrubs, and the whole stock market for that matter, can cope and seethe about their paper money.
>>
>>62937792
But do most people need the frontier models? Also, Chinese EVs aren’t as good as Tesla vehicles, but they’re way fucking cheaper and “good enough”.

But when it comes to math, medicine, advanced materials discoveries that could make buildings and rockets lighter and stronger, yeah the frontier labs own that. It could actually come down to the federal government backing them up.
>>
>>62937792
nope
frontier model usage is going consistently down
they're really more of a marketing gimmick
>>
File: will spag.mp4 (2.16 MB, 1280x712)
2.16 MB
2.16 MB MP4
>>62937859
>But do most people need the frontier models?
Yes.
Having more intelligence has always been beneficial.
Case and point: look at AI videos.
Left version from 2023 was just animated blobs. Right version in 2026 looks better than most blockbuster Hollywood movies.

If China falls behind, it means Anthropic can literally sell the "make a movie" button and destroy them.
>>
>>62937792
So you are telling me that in a year i will get what Anthropic offers for FREE?

And Anthropic wastes trillions for barely a year of progress?
>>
File: 1792179328007242.png (164 KB, 907x1082)
164 KB PNG
>>62937884
Pure cope, oh my god.
Go tell Tim Sweeney that.
The next Unreal Engine will only use AI.
I rather trust him than what basement cucks have to say.
>>
>>62937889
AI has a future, but Anthropic hasn't.
You cant compete if your competition offers the same thing for free.
>>
Xitter is full of ANT baggies who are mad that the stock isn't pumping because the solving of a millenium prize.
Inject this shit straight into my veins.
>>
>>62937917
then why is your stock crashing?

Clearly you are the one stuck in the past. You are the luddit who refuses to accept that in the future, everybody can build their own AI.
Adopt or get left behind.
You are getting left behind and lose billions every second that passes.
>>
>>62937925
China is years behind America in AI. They also can't innovate.
Every technology they have are just stolen from other countries. They are dumb as rocks.
>>
>>62937961
now it is YEARS.
Are you serious?
You are fleeing into a constructed fantasy universe.

Fact is that LLM-as-a-service is dead.
Everybody can build LLMs, everybody can access the same data that Anthropic trained on and everybody has the algorithms to do it.
There is no company secret here. There is nothing that Anthropic does that others cant.
>>
>>62937889
Most people are complaining that models aren’t getting smarter. They are in fact getting smarter, but they don’t see it because their domain has already been maxed out. They use models to draft emails and find recipes, they’re not decompiling GTA VI or doing one-man films.
>>
>>62937958
>then why is your stock crashing?
Anthropic will have another big announcement soon. It will shatter world history.

They haven't revealed their full cards yet. It's just getting started...
>>
>>62937993
>they’re not decompiling xyz
thats already maxxed out, though

Where do you see potential for improvement here?
>>
File: zijeimerehjg.jpg (296 KB, 1222x1287)
296 KB JPG
>>62937993
Most normies are retards.
The real money makers comes from the few 5,000 with internal access.
>>
Anthropic has no place in the AI future
>>
Why isn't everyone impressed by solving math puzzles?!?!?!?!?!
>>
>>62938019
why dont they just ask those mystical internal models how to stop their stock from crashing?
>>
>>62937889
>Knife spawns garlic slices
>Tomatoes disappear
>Colander spawns water
>Videos have to be slow and stiff so the model doesn't shit itself

LLMs are lossy and can never be as coherent as their inputs.
>>
File: dariocancer.png (267 KB, 1324x824)
267 KB PNG
>>62938022
Dario says they will cure cancer in 10 years.
That alone puts them ahead of every company on Earth.
>>
>>62938012
When it comes to decompiling? Doing it faster. Other domains? Math will eventually move into the machine realm, we won’t actually fully understand it all, but it’ll be verified by lean and work for new applications.

But this shit is very expensive and the federal government probably won’t have any choice but to subsidize the labs.
>>
A CNN mmmmm
>>
>>62938047
>Dario says
Dario is a low iq idiot who only got his position because he is a lying jew
>>
>>62938040
T...th....they just won't, OKAY!
>>
Heading for sub 80 KEEEEEK
>>
>>62938053
>When it comes to decompiling? Doing it faster
any local model can do that, though

Where is Anthropics place here?
>>
>>62938081
Dario is probably the smartest guy in America. You’re exposing your IQ to be near room temperature.
>>
File: sub 136.jpg (35 KB, 378x446)
35 KB JPG
Nvidia just went sub $136
>>
Anthropic is literally performing worse than crypto memecoin rugpulls WTF

Will we get a Coffeezilla investigation about this scam?
>>
>>62938104
if the inbred jew is so smart, why is his company crashing and burning?
>>
File: darioraq.png (1.67 MB, 931x1167)
1.67 MB PNG
>>62938104
They hate Dario because he was against the neocon wars.
Now look at all the anons in this thread sucking China's dick.
They can't stand America winning.
>>
>>62938144
i am sure the jew was against wars for jews, totally, absolutely
>>
>>62938144
you know who isn't winning?
Anthropic shareholders and investors
>>
IT WILL BE BELOW 80 AT MARKET CLOSE LMAO LOOOOOLLL

The fucking Melania Trump memecoin performed better than this
>>
>>62938047
you know there's a cancer from a woman that died 70 years ago still living and reproducing in labs all over the world, often contaminating other tissue samples

what would it look like if it took on a human form?
>>
File: IMG_4429.jpg (28 KB, 218x265)
28 KB JPG
>>62938134
>this company founded five years ago is now only worth only $1.2 trillion
Are we having a serious discussion here? Go back to sniffing sharpies.
>>
>>62938269
who gave them money during those five years?
>>
>>62938293
governments horny to have a working crystal ball, something they've been researching for the past straight century
>>
>>62938269
you do know that they are 900 billion in debts?
And the stock value will go below this, which means that shareholders are basically just their lenders who trust that the government will pay it out somewhow.

You dont invest in something with value, you give jews money in the hope that jews can use other jews to make taxpayers give it to you.
>>
>90% of shares are still privately held waiting to be dumped
So bullish ANT
>>
>>62938321
>working
Isn't the Iran war run with Anthropic AI predictions?
Didn't the FAA have the worst system crash in their history a day after deploying Anthropic vibe coded shit?

What is their track record? Can you point to a single Claude created project that wasnt a failure?
>>
File: IMG_4271.gif (72 KB, 498x354)
72 KB GIF
>>62938348
>AI isn’t valuable
Next few years is going to be very rough for you
>>
>>62938407
that is not what is written there, anon

Anthropic is not valuable. AI can exist without an LLM-as-a-service construct.
>>
>>62938397
>what is Claude Code
You know, that thing that has almost automated software engineering
>>
>>62938425
>what is Claude Code
the cause of the largest FAA system crash to date
>>
>>62938397
i didn't say it actually worked, i said it gets government prop-ups because they think it'll eventually work
>>
>>62938440
This is speculation.
>>
>>62938425
go ahead, sell software built with claude what could go wrong
>>
File: 1783538491073566.jpg (124 KB, 1080x593)
124 KB JPG
>>62938040
>>
>>62938425
Name a single company that successfully deployed Claude Code to create better software.
Is Windows better now than a year ago?

We have countless of examples of Claude producing crashes and slop, even if speculatory >>62938513
We have the proton CEO announcing that he finds it ridiculous that all his employees burn AI tokens, yet the software doesnt improve.
But we have no example of Claude producing something valuable.
>>
>>62938571
Who said anything about better? It’s about efficiency.
>>
File: 1769252453809.jpg (218 KB, 779x778)
218 KB JPG
>>62938571
We even have scientific studies saying that Claude increases code output, but not the output of actual software.
Companies who use Anthropic, do not experience a rise in quality of efficiency.

Thats why Anthropic has no future.
AI may stay. AI may be very useful. But Anthropic is not.
Adapt or get left behind. Anthropic, as a company, is a construct of the past.
>>
>>62938602
Can you name a company whos efficiency has improved thanks to Claude?
>>
File: steam.png (146 KB, 1521x1356)
146 KB PNG
>>62938540
>sell software built with claude what could go wrong
There are now hundreds of AI-assisted video games being released on steam every day.

Seems to be doing fine. :)
>>
>>62938571
These people don't realize that you still have to be a software engineer to actually get a usable product from Claude Code. Claude Code isn't bad at all, but I am a software engineer and I know how to use it. I've also used DeepSeek, Codex, etc and still got good results.
COBOL was literally made so that business people could replace their programmers. Guess how that turned out. AI is not good at making decisions - you aren't going to make the next NT kernel without years of OS theory.
>>
>>62938540
you mean every single modern operating system?
>>
>>62938602
efficiency means a worse product for cheaper. In your eyes.
>>
>>62938695
so you are saying that Windows is worse now, than it was a year ago, because of Claude?
>>
>just hit 83
So... what is the floor?
>>
>>62938711
>for cheaper
throwing trillions into chatbots doesnt sound very cheap either, desu
>>
>>62938719
Windows is just as shitty now as it was a year ago, except now we can finally move the taskbar again. It's better now, which doesn't mean much.
>>
>>62938734
below 10 probably

But it will take years to get there
>>
File: IMG_0601.jpg (266 KB, 1750x2500)
266 KB JPG
>>62938627
Anthropic has no future because maybe just maybe it was founded by a bunch of jewish thieves swindlers and tricksters. The only logical conclusion any man of sound mind and intelligence should conclude going forward is total kike death and nationalizing and seizing all jewish assets. This time we can burn the demons in nuclear reactors
>>
>>62938571
It does have uses. You should however assume that the actual uses lag at least six to twelve months to what is claimed by the AI influencers and the labs.
>>
>>62938768
didn't we just have an issue of nvidia drivers writing terabytes of data to your ssd?
>>
>>62938746
So a massive amount of money for a worse product? (clippy)
>>
File: carl finger.jpg (29 KB, 400x400)
29 KB JPG
>>62938770
Keep dreaming kid.
>>
ALMOST 82
LMAOOOOOOOOO
>>
>>62938571
>Name a single company that successfully deployed Claude Code to create better software.
>Is Windows better now than a year ago?
This. Microsoft CLAIMS that it's finding thousands of bugs a month using AI, but when you ignore the hype and dig into the details, you find that the agents are using pre-existing, non-AI tools (static analyzers, fuzzers, etc.) that Microsoft didn't bother putting in the work to write domain-specific glue code for until it had a financial incentive to overhype AI.
>>
Insiders dumping on yall like no tomorrow. Thank god im short
>>
>>62938269
>worth only $1.2 trillion
It is at 1.1 trillion now.
Yesterday at this time it was 1.6 trillion.
>>
>>62938571
>my software runs 30% faster
>the end user not caring that now he has 5 secs more of free time
I have yet to see ai bring anything tangible for the end user.
>>
>>62938848
And originally vamued themselves as 2 trillion.
>>
>>62938793
>nvidia drivers
not Windows
>>
>>62938960
nvidia is literally an Anthropic provider and investor, though
>>
>>62938571
>But we have no example of Claude producing something valuable.
The only valuable thing AI agents have made is custom content companies refused to. Like reverse engineering of expensive software, old/console games, making mods that are actually good (unlike western AAA games).
But claude blocked this now so that revenue is going to chinese models
>>
File: 1747657867421.jpg (51 KB, 472x591)
51 KB JPG
>"Ugh, this Apple company is 90 days from bankruptcy. Their stock is worthless!"
>"Steve Jobs? There's no way he can turn things around."
>>
>>62937837

all that cash is investor's money, not earned profit
>>
>>62939004
Apple didn't crash 40% the first two days after IPO
>>
should have bought monkey jpeg NFTs instead
>>
>>62938734
With 30 billion in revenue, around 90-120 billion mcap
>>
>>62938602
ok so software shrinkflation got it

go back to skool FAGGOT
>>
>>62939120

$8-9 per share then. It will be long trip down.
>>
File: wew.png (57 KB, 706x533)
57 KB PNG
>>62936798
the market decided
>>
File: 1782720558989377.png (110 KB, 258x245)
110 KB PNG
>sit on riemann hypothesis
>use it to try and pump IPO
>lose hundreds of billions
LMAO
>>
>>62937503
KEK my first thought too. If math isnt serving a productive purpose, it's useless.
>>
File: marathon nov 20.png (177 KB, 1129x259)
177 KB PNG
Now that the dust has settled, was Anthropic the Marathon of stocks?
>>
File: 1790284307357517.jpg (119 KB, 400x400)
119 KB JPG
>>62937604
>P.S professors are mad because the AI can't show it's working. An understanding of how the math was done is extremely valuable so that it can be built upon
Good. They did it to science first, its only fair (computer) science does it to them in kind
>>
So its worthless now
>>
>>62939492
Comp Sci was always the superior science anyways
>>
>>62937792
>impossible
Time will tell. They do seem to be suffering now though, and Nvidia chips being cut off wont help them either. I think its also inevitable now that the US gov will put up regulations againt companies using chinese models. Either hard bans, refusing to directly work with companies, or like recently rumors about forcing them to disclose they use chinese models like they are some steam released video game kek.
Google (and maybe Meta) seems like the bigger risk to Anthropic long run, but for now Anthropic seems to have succeeded in scooping up the most AI related talent. Though the tech is new, who knows what direction some new discover/breakthrough takes this stuff

>>62938040
Use case for caring about your stock crashing after you sold? Seems like they timed things near perfectly
>>
>>62939599
Top AI talent are sitting at home laboring inj the factories actually on the front lines building local not working for jewish tricksters and swindlers
>>
>>62936798
Bullish, just boughted 20 shares for 86.2$
annoyed I couldn't get the 83.5$ shares but I wasn't logged in for that.
I'm very long-term on these though, and expect to lose some in an overall market correction.
>>
>>62936798
Fuck dario
>>
>>62939599
> its also inevitable now that the US gov will put up regulations againt companies using chinese models.
Maybe, but I doubt American allies will follow
>>
>>62937588
I mean anthropic will be profitable if trend continue. The made 40 billion this year up from hundreds of millions up from a few millions. Unless everyone moves onto another competitor or local they could easily catch up to their debts since the bulk of their startup is done.
>>
>>62936798
You really think any tech stock is going anywhere other than down right now? The oligarchs pushed too hard and now they're getting hit with a big dick dem deepening of their anuses.
>>
File: extrapolating (1).png (21 KB, 461x295)
21 KB PNG
>>62939712
>I'm very long-term on these though
Funny how you're suddenly "long term" and not "going to the moon soon" when you're obviously losing.
>expect to lose some
Then why buy? Sell and come back in a few weeks when it's <$50.
>>62939767
Pic rel.
They can only ever sell the most expensive to make high end models and right now google beat them there too.
>>
>>6293976740
40 billion does nothing at all to justify their market cap. It doesn't even put a dent in their capital expenses, and VC loan liabilities.
>>
>>62938571
>We have countless of examples of Claude producing crashes and slop
surly you can provide one then
make sure to show the formal verification used as well
>>
>>62939492
>They did it to science first
what?
>>
I boughted over 100. Its going back up right?
>>
>>62939735
I used to think that too. But keep in mind Trump is seen as anti-regulation on AI and most western countries get hard from the thought of over-regulating things in general, those two things will be enough for a lot of western leaders I think. I know for here in Canada, shortly after Dario's pace the frontier blog post Carney went and signed some commitment with I think Finland and Germany (and other countries no one gives a shit about) to pace the frontier. Granted it seemed a toothless agreement but its the direction the leaders seem to be looking. Cohere the Canadian AI company also recently released some information claiming Qwen (chinese mode) is trying to send into back to china.
Counties without AI companies seem to see pacing the frontier as a chance to not fall as behind and ones with AI companies see it as a way to boot out open weight chinese competition.
(Also the EAs are 100% paying off politicians throughout the west)
>>
>>62939888
Checked
No, hope you don't kill yourself anon
>>
>>62939888
It'a okay anon, you were manipulated for the selfish gain of some of the most callous people in the world. One day you'll look back and realise the lesson you learned about how people can work was worth 1000x what they took from you
>>
>>62938003
>>
>>62939904
Oh fuck you might be right...
>>
Only got allocated 25 shares in the IPO when I requested 150 so I took the opportunity today to buy another 125. Claude is my friend.
>>
File: vampire.jpg (9 KB, 319x180)
9 KB JPG
They are the best right now when it comes to AI agentic coding, but there are like 3 other companies that are pretty close and some have better chatbots.
I don't really get why it's rated that highly. They could lose the lead at point in the near future.
>>
>>62939015
Investors won't really get to decide how it's used. Founders have 50.1% guaranteed voting power.
>>
Math proofs are relatively easy for these models.
If only it was as easy to prompt for breakthrough drugs. The company might be worth something.
>>
>>62940311
Just gotta get the right RL environments set up to hill climb. t. computational chemist
>>
>>62940221
coding is the primary usecase of LLMs. nobody cares about pure math and every other application so far is either too easy or grafted-on cope that isn’t worth the squeeze
>>
>>62940221
it's because coding is such an inflated field that a small edge in that is worth billions, but we'll see how long it lasts.
>>
haha
hahaha...
HAHAHAHA...
>>
File: 1787141405186261.jpg (22 KB, 618x455)
22 KB JPG
>https://www.reuters.com/markets/us/anthropics-ipo-bank-closed-position-meant-stabilize-stock-removing-source-2026-11-20/
ahahahahahahahahahaha
>>
>>62937599
2 weeks
>>
>>62944091
>Anthropic named investment bank Morgan Stanley as the initial public offering's stabilization agent to support the stock after its market debut. In that role, the bank received an option, known as a greenshoe, to buy additional shares — 147 million in Anthropic’s case — directly from the company on behalf of the underwriting syndicate.
So the two times it was at 83.x and got a slight boost up again, it was actually Anthropics hired agents who bumped the stock?
This is a larger clusterfuck than anyone could have dreamed of.
>>
>>62945572
I don't think so. I think they were dumping everything they could and as soon as the price approached $83 they stopped selling and let baggies load up a bit so they could get dumped on some more.
>>
>>62940221
>They are the best right now when it comes to AI agentic coding
No, ChatGPT is the best currently for coding in all benchmarks.
Anthropic is the most shilled and they pour the most into spambots to make you fall for their lower quality product, though. They even had a public website with shilling instructions, that they shut down when it got found.

Just look at their shills here. They tell you that free models are a year behind, and when you question how useful it is, to have your product offered for free in a year by others, they upgrade it and claim that it is years.
It's all just lies.

I recently had a problem. Specifically it was about skinning matrices for 3D rendering, where Claude completely shit the bed and i asked Deepseek and he got it right.... for free. There are many different aspects to coding and in reality you will find one case where x is better and another one where it is y.

Anthropic does nothing special.
Anthropic could cease to exist today and nothing would change.
Vibe coders would still produce the same low quality of slop, just with a different model, AI would still be progressing... just without Anthropic.
>>
>>62945589
Sounds like bullshit to me. I never even used Claude but from what I've seen the notion that it can't do something as basic as skinning matrices is absurd.
>>
>>62945611
>i just don't believe it
you do you, anon

Claude Code created Artcraft has thousands of known bugs. Funny stuff like "when zoomed in, Ctrl-Z Undo doesn't work". It is all things that make you wonder how the fuck this is even possible.
>>
>>62945641
It's not that I just don't believe it, they are taking in billions in revenue and I can ask google AI to write a skinning matrix shader and it will do it perfectly every time. So it's not that I don't believe it, it just doesn't add up logically.
>>
>>62945647
Skinning matrices are handled different by engine and format.
One file format has bind shape matrices, another one doesn't. Some engines restrict what you can do with joints. Maya can have a neutral scale applied in its rest pose that gets baked into the bind shape matrix, Blender doesn't. Some engines even only have one skeleton that gets morphed for each entity.
Baking against a different rest pose is also irreversible. If you have one mesh that should fit on top of another while a common armature rigs both, and your AI writes 20k LoC for baking shit, they will clash.

If you look at meshes created by AI, you will notice that they all hit a wall when it's about skinning.
>>
>>62945691
Man just sounds like you are trying to flex your rigging experience. So Maya adds a single scale matrix on the right side of the composition? you aren't going to flim flam me in to believing that AI can't figure that out.
>>
>>62945724
When there are 100 different established ways of doing something and you ask an AI that references all of them, what do you end up with?
We are talking about a model that made software dump tens of TB of data to your SSD per day.
>>
>>62937345
Most boring technology in the human history.
At least trains are exciting to the autistic.
This just bruteforces the world's most boring tasks (with heavy human scrutiny and oversight needed) so it can code the most basic shit imaginable or help solve the pee pee poo poo hypothesis.
>>
>>62945756
Cope. AI is interesting to autists.

Ai = autistically interesting
>>
File: 1790665285085551.mp4 (3.9 MB, 1080x1920)
3.9 MB
3.9 MB MP4
>>62945756
This is the moving goalposts I was talking about.
AI has helped improved video gaming modding by x1000 and soon it will generate any video game imaginable.
>>
>>62945848
Cool, but it isn't profitable
Also, coding with Claude is infinitely less interesting and yet everyone is forced to do it
>>
>>62945896
>but it isn't profitable
Enabling scaling labour up on demand isn't profitable?
>>
>>62945896
There's just no reasoning with you people anymore.
You will plug your ears and deny every evidence thrown at you.

I just wish you bastards would stop crying when you get fired from your jobs.
I rather have an AI coworker at this point.
>>
>>62945901
It is not that profitable if what you offer is a commodity with no lock in. How profitable is renting construction equipment? It is enormously manual labor saving.
If you're the only company renting good construction equipment or can build something patentable, then it's very profitable.
>>
File: 1795218720823497.jpg (435 KB, 1536x1024)
435 KB JPG
>>
>>62937345
All of these proofs are just hacks that exploit bugs in Lean to make it "verify" complete nonsense
>>
>>62945989
What do you think will happen once AGI is achieved, anon? You think governments around the world will just allow any company to create intelligence on demand?
No, they will allow specific companies create a specific amount of bodiless minds per year to avoid completely destroying the economy.
Anthropic will become (part of) a legal monopoly, endorsed by the US government and accepted by lesser governments.
>>
>>62945989
>It is not that profitable if what you offer is a commodity with no lock in.
I've used to AI to my cut my art budget in half at my business.
We don't need graphic designers, video editors, or 3D artists anymore.
Claude, ChatGPT, Nano Banana and a slew of other tools can create anything I need without question.

The money I've saved is now counted as "profit".
A 3D artist for example, would demand $50 an hour. He's gone and I saved $1000s because of it.
>>
File: 1770034479026904.jpg (5 KB, 250x250)
5 KB JPG
>>62937345
OUR STOCK IS DROPPING
QUICK, DEPLOY THE RIEMANN HYPOTHESIS
>>
>>62946007
>The money I've saved is now counted as "profit".
And you are the only one to do this and can keep these profits in perpetuity?
>>
>>62946022
>And you are the only one to do this and can keep these profits in perpetuity?
Dude, there are literally people on this board who swear they'll never touch AI or refuse to embrace it as a tool.

I will keep profiting as long as those dumbfucks never do their research. And that's fine by me.
>>
File: ai medal.png (59 KB, 1528x647)
59 KB PNG
>>62946002
So why did AI already win at the Math olympics as old as 2024?
https://deepmind.google/blog/ai-solves-imo-problems-at-silver-medal-level/

Take Problem 6 of the 2024 International Mathematical Olympiad. Only five contestants solved it. Humans translated the problem into Lean, AlphaProof produced a solution, and mathematicians Timothy Gowers and Joseph Myers assessed it for full marks. Anyone claiming that result was a hack needs to explain what went wrong in a specific solution that mathematicians could inspect.
>>
>>62946004
If they can capture goverment. There is no reason for the state in your scenario to allow these legal monopolies to extract massive rents. At least not domestically.
>>
File: dimmadome.jpg (74 KB, 500x627)
74 KB JPG
>>62946041
>There is no reason for the state in your scenario to allow these legal monopolies to extract massive rents.
Considering we've had governments act against their own interest (and against the interest of the people they're supposed to represent) for many decades, largely due to lobbies funded by corporations, I'm just going to label you as a retard and will be ignoring your posts from now on.
>>
File: 8888.png (40 KB, 497x363)
40 KB PNG
By the way, the doomers are running out of time.
There are now 4 millenium math problems left.
If Anthropic solves all of them by Christmas, the world as we know it will permanently change.
You have been warned.
>>
File: dario.webm (782 KB, 396x402)
782 KB
782 KB WEBM
>>62946060
Which political faction is going to elevate this guy to a god king by law?
>>
>>62946069
>spend 30 million in tokens
>receive 1 million for solving the problem
Yep, sounds like something right up Antrophic's alley.
>>
>>62946007
>>62946032
>>62946038
>>62946069
two more weeks of unverified claims and fake CVEs and it will stop the stocks from crashing for real this time!
>>
>>62946079
If they solve P=NP they will own the entire world economy.
AI could defeat every encryption and take down the entire internet.

It's Hiroshima and Nagasaki on steroids.
>>
>>62946099
>we will cause damage and get jailed for it
how does this generate profit?
>>
>>62946103
What jail? You think any nation could invade the USA?
They'll get bombed to oblivion.
>>
>>62946113
so you only take down "the whole internet" within the USA?
>>
>>62946124
Man you're stupid.
>>
>>62946099
Splurting out a 6 gorillion page "proof" is very different from actually making a working algorithm that can defeat real encryption.
Also I really don't see how taking down the entire internet would generate profit for AI companies, they're pretty reliant on people having internet access.
>>
>>62946132
Isn't the whole Western internet already taken down by AI? Like all Western platforms, especially from the USA, are flooded to death by spambots thanks to companies like Anthropic.
>>
>>62946136
>Also I really don't see how taking down the entire internet would generate profit for AI companies,
Lets see how long the rest of the world (i.e outside USA) would survive when all their infrastructure is now offline?
By the second day I guarantee you'll be eating dogs off the street.
>>
>>62946159
>Lets see how long the rest of the world (i.e outside USA) would survive when all their infrastructure is now offline
But how does this make money?! Where's the profit? Wrecking random shit and making people poor might be funny but it's not a 2 trillion business.
Also the US could already do this if it really wanted to, basically all of Europe, South America, and Japan rely on American cloud giants for the majority of "their" internet.
>>
File: a10-shot-down.mp4 (356 KB, 640x400)
356 KB
356 KB MP4
>>62946159
ok, zogbot
>>
>>62946145
Then it shouldn't bother you when Anthropic continues to release more models and replace jobs.
Personally I don't care because all the rich people will just move towards their own communities for real human contact again.
>>
>>62946191
that does indeed not bother me

The only thing Anthropic does, that affects me, is running spambots and making our Western internet miserable.
It also downed all planes of the USA for half a day, but i didn't fly at that time, so that didn't affect me.
>>
>>62936999
trips of truth
>>
>>62946178
>But how does this make money?! Where's the profit? Wrecking random shit and making people poor might be funny but it's not a 2 trillion business.
You just answered your own question.
It means the rest of the world is now permanently poor and a more powerful country can take whatever they want next.
>>
>>62946197
The industrial revolution also had a lot of growing pains but the end result gave us new cars and clothes that didn't take an entire year to make.
The AI transition is messy right now but it's worth it when I get my own maid robot that can cure cancer and create movies on demand.

So I rather invest now and secure myself a better future. For those against it, lol good luck.
>>
>>62946201
Would you rather have a wealthy European country as their vassal or an African shithole?
Which one will benefit you more?

What you don't seem to understand is that rulers have a better and easier time and are more wealthy, if the people who they rule over do well.
Rich people do not want to hide in a bunker and be in a constant state of fear, because any random pleb on the street could run them over and kill them out of spite.

The theory, that making people miserable and spreading death is profitable, is wrong.
How much did we gain from the Iran war? Did it make us more wealthy?

Meanwhile China doesn't go to war and doesn't take and milk vassals, and they are doing better every single day.
>>
>>62937345
AGI IN 2 MORE WEEKS
DON'T BE A PANICAN TRUST THE PLAN
>>
>>62937345
A RIEMANN HYPOTHESIS SOLUTION JUST FLEW OVER MY HOUSE
>>
>>62946220
>Would you rather have a wealthy European country as their vassal or an African shithole? Which one will benefit you more?
Who said anything about a wealthy Europe?
They don't have a choice and can't dictate anything. If they're just as poor as Africa that's their failure for not competing.

>What you don't seem to understand is that rulers have a better and easier time and are more wealthy, if the people who they rule over do well.
That was before AI and Robots exist.
When robots can do everything the average human literally serves no purpose anymore.

>The theory, that making people miserable and spreading death is profitable, is wrong.
Capitalism by design requires inequality.
You don't reward the smartest person with a bad salary. You pay them more because they can do more than a lazy person.

If it took you that long to understand then you are truly naive about how the world works.
AI is just meritocracy on steroids.
>>
>>62946079
lmao
>>
>>62946201
The US could've done this at almost any point past ww2. And they didn't not because they wanted to be good guys but because the system of vassals and global dollar reliance was FAR more profitable. It's better to take a skim off the top of wealthy economies than it is to totally rule over a poor shithole.
>>
>>62946060
Reddit commie slop. This is a board for making money faggot.
>>
>>62946249
The President of the United States of America has to board planes inside luggage containers, because he is scared about the possibility of some random person with a drone blowing him up.
The main usecase of AI so far is spaming on the Western internet and Palantir mass surveilance in the West.

Do you think that this is a good business model? To make your own people miserable and to have your leaders hiding in suitcases in fear of getting their head blown off?
>>
File: ezgif-860494e64446b49e.jpg (383 KB, 1600x1200)
383 KB JPG
>>62946263
Did you intentionally forget about the Cold War?
Right after WW2 the Soviets had the biggest army and they were propping up Communist governments worldwide.
The U.S could have ignored them but there would have been far enemies on her border.

But it's not 1991 anymore. Every country can go their own way now.
>>
File: ezgif-359b488bd10276f1.jpg (1 MB, 1200x716)
1 MB JPG
>>62946272
>The main usecase of AI so far is spaming on the Western internet and Palantir mass surveilance in the West.
Explain how making robots and putting them to work in factories is "spamming"?

This is still AI. Amazon has been using more robots to get better efficiency from their warehouses.
We can scale this up and do this for every job.
We don't need humans working when a machine can do it faster and cheaper.
>>
>>62946264
I completely forgot that you make money by only buying stocks after they've shot up to the moon. My bad!
On a more serious note, if you don't expect this stock to rise back to its starting point within the next 6 months, you're not gonna make it.
>>
>>62946304
We have robots in factories since 1961, anon.

And your point was that AI will kill the internet and make everyone miserable and that this is somehow super profitable because having a population that is edging on Luiging all rich dudes and leaders is somehow great.

You are just an idiot, who thinks that it sounds cool to be evil.
And now that your position got wrecked, you go on: "AI will do what we already do for 65 years now".
>AI Amazon
nigger, they are putting boxes from A to B, that is far less impressive than what robots are doing in car manufacturing plants for decades now already
You only think that shipping boxes is a huge feat, because you are living in an "economy" that produces nothing. You would have gotten your mind blown if you would have visited a factory that produces pencils twenty years ago and the state of automation.
>>
>>62936798
>>
File: file.png (36 KB, 649x202)
36 KB PNG
the goyim got fleeced again
>>
>>62946335
>We have robots in factories since 1961, anon.
Those 1961 era robots are far too primitive compared to what 2026 robots are now doing.
In 1961 a robot could only unload a die-casting press by repeating a sequence of movements stored on a magnetic drum.
In 2026, you could ask a robot to ‘pick up the bag about to fall off the table" without anymore instructions. The robot can use its eyes to identify objects just by names alone. A worker could also describe any problem in plain english and the 2026 robot will understand it automatically.

>And your point was that AI will kill the internet and make everyone miserable and that this is somehow super profitable because having a population that is edging on Luiging all rich dudes and leaders is somehow great.
Well no, the internet would still exist. It would just be replaced by more tighter knit communities if human contact is your thing.

>You are just an idiot, who thinks that it sounds cool to be evil.
You are a luddite who stands in the way of progress.

>nigger, they are putting boxes from A to B, that is far less impressive than what robots are doing in car manufacturing plants for decades now already
Amazon currently employs around 1500 to 2000 people per warehouse. If humans are just "putting boxes from A to B" then I support robots taking over those jobs.

>You would have gotten your mind blown if you would have visited a factory that produces pencils twenty years ago and the state of automation.
Cool, that still doesn't change I want all factories to be 100% automated.
>>
>>62946395
Here you got a factory producing pencils 20 years ago:
https://www.youtube.com/watch?v=Zsnw98aTWOM

Barely any human around, 99% of the workforce already gone due to automation for decades already. It already removed hundreds of jobs there. The only thing it can still remove is some dude with a forklift and some woman checking for damages.
So two jobs?

And what is the role of Anthropic in this?
Do i need 2 Trillion USD Anthropic to detect when a pencil is damaged? Probably not.
Do you think it is more economic to upload a steady stream into the cloud to Anthropics AI-as-a-service, or wouldn't it be better to train one locally?
>>
File: 7777.png (1.12 MB, 1107x613)
1.12 MB PNG
>>62946342
Imagine being upset in 2026 we have companies willing to advanced math and scientific research for the greater of humanity?

Better than spending it on a gay parade.
>>
File: 1792765862954058.webm (2.3 MB, 1150x670)
2.3 MB
2.3 MB WEBM
>>62936798
long term bullish
short term bearish
>>
File: 1775149874473187.png (622 KB, 748x408)
622 KB PNG
bullish by december 1st
>>
File: ezgif-7cf3963d08298a28.jpg (909 KB, 3125x3125)
909 KB JPG
>>62946421
Nigga, not every company works on pencils.
You posting an automated pencil factory doesn't refute the HUMAN factories that employ 1000+ people.
And what's Anthropic role? They are providing the software that can either directly control robots or enable them to work efficiently in those warehouse environments.

It seems odd why you are trying to fight or disagree with me on this?
>>
>>62946219
its annoying to think i was born too early to enjoy the robot easy life. just in time to see it take shape.
>>
>>62946429
>closed source
>(((ethical altruists)))
>"le greater of le humanity"
fuck off bugperson
>>
>>62946469
Making it open source wouldn't stop the people who already own render farms and can work on solving these equations that would make your PC chug.

Again, you guys need to take the L if you really believe solving math is now evil. LOL.
>>
>>62946443
>not every company works on pencils.
It is a more complex task than moving a box from A to B

>Anthropic is providing the software
They don't. The software to train AI models is open source, the algorithms are know. Every neet can do it. It's a markov chain.
Anthropic has absolutely nothing that we can't access for free. They don't even have any special know-how. They have no purpose here.

Anthropics business model is AI-as-a-service.
Do you think that Amazon warehouse robots livestream their data to Anthropics AI-as-a-service to control them? Nope.
Why would any factory, that uses AI for automation, stream their shit over the internet to have an Anthrpic slopbot decide, if they can run it locally?
>>
>>62946527
>It is a more complex task than moving a box from A to B
Wow, you're getting desperate now.
There's more to factory life than just moving a box.
Hell, if liquid spills on the floor you need to pull someone off the floor to get a mop and clean it up. Your stupid pencil factory never achieved that.

>They don't. The software to train AI models is open source, the algorithms are know.
There is no open source model that's as powerful or as capable as Claude.
My claim is that Anthropic’s software can help people build and operate robotic systems, and we can test that claim directly.

Anthropic actually ran that test. It split eight researchers into two teams and asked them to program robot dogs to fetch a ball. One team had Claude. The team with Claude completed more tasks and averaged about half the time on the tasks it finished. This was a small company-run experiment, so it establishes a narrower and more interesting point: a model can help write and debug robot software even when the finished robot runs locally. A factory could use Claude during development and disconnect it before the first production shift.

>“Anyone can train a model”
This skips the expensive part. Anyone with suitable hardware can train a model. Reproducing Claude means obtaining comparable data, computing power, and training results. Anthropic documents using nonpublic third-party data, paid labeling, and internally generated examples. An open-source training program gives you none of those for free.

> It's a markov chain.
If you define the entire conversation so far as the current state, you can describe next-token generation that way. That mathematical description tells you remarkably little about what the trained model can do. The practical question is whether its output helps someone get a robot working. In the experiment above, that’s exactly what the researchers measured.
>>
>>62946587
How much money did Anthropic earn on Amazons warehouse robots?

I looked it up, and it is a total sum of $ 0.00

Nothing. Anthropic has no role here.
Don't be a Luddit. Accept that AI-as-a-service is an outdated dead business model.
Adopt or get left behind. In case of Anthropic, they are literally getting left behind.

If you want automation, you want LOCAL models.
>>
>>62946601
No one wants your local shit.

Here's a fun challenge for you: use your "local" mode and try and create GTA 5 or 6 with a prompt?
Can't do it? Because the next Anthropic model in 2027 can.
>>
>>62946616
Money Anthropic earned on Amazons warehouse robots, who run LOCAL models: $ 0.00

Your prime example for what AI can do, is a local model.
>>
>>62946355
Tra le ipotesi c'è appunto anche quella di una breve vacanza all'interno di un resort. Mentre nell'agenda del Ceo non ci sarebbero uscite in yacht nell'area del Brindisino. Ma c’è anche chi lega la visita a motivi di affari. Tanto più che Brindisi è stato teatro in questi giorni di misteriosi movimenti internazionali.

Lo scorso 17 novembre è stato avvistato in città il ministro della Difesa della Repubblica Popolare Cinese, l'ammiraglio Dong Jun. Il ministro è stato ospite della Marina Militare italiana. Si è parlato di un viaggio istituzionale in Italia, sul quale c'è stato il massimo riserbo.

L'incontro è avvenuto a bordo della nave Atlante ormeggiata al porto di Brindisi dove Dong Jun è stato accolto dai vertici della Marina Militare. Durante la visita i fucilieri della Brigata San Marco hanno eseguito un'esercitazione dimostrativa nelle operazioni anfibie e nella sicurezza marittima
Among the various theories is the possibility of a short vacation at a resort. Conversely, there is no indication in the CEO’s schedule of any yacht excursions in the Brindisi area. Some, however, link the visit to business matters—all the more so given that Brindisi has recently been the scene of mysterious international activity.

On November 17, the Minister of Defense of the People's Republic of China, Admiral Dong Jun, was spotted in the city. The Minister was a guest of the Italian Navy; the visit was described as an official trip to Italy, handled with the utmost discretion.

The meeting took place aboard the vessel *Atlante*, moored in the port of Brindisi, where Dong Jun was welcomed by the Italian Navy’s senior leadership. During the visit, marines from the San Marco Brigade conducted a demonstration exercise focused on amphibious operations and maritime security.
>>
File: 1794807814238064.jpg (271 KB, 1206x1934)
271 KB JPG
>>
>>62946624
https://bari.repubblica.it/cronaca/2026/11/20/news/il_ceo_di_anthropic_dario_amodei_atterra_in_puglia_mistero_sulla_visita_del_guru_dell_ia-425737920/
>>
File: hqdefault.jpg (20 KB, 480x360)
20 KB JPG
>>62946629
>>62946624
Hey doofus, this is an English website.
>>
>>62946627
kek
>>
>>
>>62946664
Wow, China sounds like a great place to live!
So when are you packing your bags and moving there?
Don't tell me you're scared of their police state tactics where you can be jailed for anything?
>>
>>62946738
Exactly. I'd rather have flock cameras on every block and a federal israeli agent showing up to my house after posting something online like in the US of A.
>>
>>62946752
>and a federal israeli agent showing up to my house
This part is imagined btw
>>
File: bane.gif (970 KB, 245x152)
970 KB GIF
>>62946752
And you think there isn't a large list of things that would cause the police to show up on your door if you posted them in China?
>>
>>62946738
what has this to do with the outdated business model of Anthropic?
>>
>>62946762
Is it? Why are so many israel flags flown in your representatives offices'?
>>
>>62946771
So you are like china so therefore you are winning? Fuck me. I was hoping the people of the USA would have more spirit. But obviously that hope was misplaced.
>>
>>62946785
Because they control from the top, you dumbass
Why would they waste time and send their own people to people's doors when they can just modify the system and make people do what they want through tricking them into thinking it's what they themselves want?
>>
>>62945583
Jews are diabolical people. Anyone not seeing this nor getting informed about their schemes in the golden age of the information ia a retard
>>
>>62946809
Hmmm. Starting to sound like socialism... no?

Could you explain the differences between capitalism, socialism, and communism for the class please?
>>
>>62946838
Do your own homework, nigger
>>
Jews seem to be incredibly marxist. I wonder why. Seems weird.
>>
>>62946842
Oh I already know. I'm asking the question because I don't actually believe you know. But you certainly speak as if you do.
>>
90% of stocks still locked, they will be unlocked in 6 months. Guess what happens to price when the supply does 10x with the same amout of demand?

If this was a shitcoin, people would be screaming premined scam.
>>
>>62937521
>spending billions to make millions

Sounds super sustainable
>>
File: Untitled.png (290 KB, 948x1094)
290 KB PNG
>>62947026
Please continue with the funny predictions. It will look good in the screencap.

Every anti-ai prediction has failed.
>>
>>62947452
are these screencaps in the room with us right now?
>>
>>62937754
deepseek has the EXACT SAME restrictions as claude does
you have to ask Gab or a local open weight model to circumvent the safety autism
>>
>>62947805
they're in his computer
are you just repeating shit with no context? lol
>>
File: breddit fail.png (143 KB, 1080x387)
143 KB PNG
>>62947805
3 years ago the anti-ai team said robots could never make videos. It was "totally impossible" they said.

Anthropic would go onto completely obliterate such idiot.
>>
>>62947452
An AI company being good does not follow from the statement "AI is a useful technology". The publicly traded stock supply is doing a 10x in a couple of months, and that's factual reality.
>>
>>62948014
>An AI company being good does not follow from the statement "AI is a useful technology".
AI has completely replaced coders.
Literally 1 year ago that idea was seen being completely sci-fi or "impractical".
Now it's reality.

People in this thread don't know shit about stocks and what's coming next.
2027 will see a major exponential growth in Anthropic's profits.
I'm just glad I got in early.
>>
>>62948014
Whoops forgot the attachment.
It's a hockey stick in terms of growth and market capture.

And it will never slow down.
>>
>>62947999
Anthropic has never built an image model or a video model.
>>
>>62948137
It can write code and generate videos via Blender.
>>
>>62948124
That is a pitiful revenue relative to market cap. A healthy valuation is like 3-4x yearly revenue. Anthropic market cap is like 40x their revenue, which indicates extreme overpricing.

They very clearly gamed the IPO to maximize shareholder gain by constricting supply, running ads, and trying to spin a narrative around AI being le godlike tech (if AI is a genie, why are you unloading your shares lmao?), trying to end up in the top 100 so pension funds would further pump them.

The company has no moat and their tokens are overpriced as fuck. Consumer grade GPUs can run similar models just fine just slower, which isn't an issue since consumer GPUs are idle most of the time and electricity is quite cheap. Who cares if my instance of the model codes slower when I can leave it on 24/7 while going via Anthropic would gate me due to pricing?

Their only current play is trying to create a regulatory moat against open models, and betting on that is just foolish. They're currently screeching on twitter about how evil open models are lmao.
>>
>>62948168
>Company which grows revenue by more than an OoM a year should be priced at 4x annual revenue
Explain why.
>>
>>62948184
They did a 10x from 2025 because that's when coding agents started being viable, the growth % is fast at the emergence of a technology/market. It's not realistically keeping that up and new threats are emerging.
>>
>>62948168
>Consumer grade GPUs can run similar models just fine just slower, which isn't an issue since consumer GPUs are idle most of the time and electricity is quite cheap.
This has been debunked.
First, who do you think is training your local models? When is the next version of it being released?
You can't answer it because local models are completely handicapped by lack of data, compute power and training results.

Anthropic does not have that problem. That's their "moat".
>>
>>62948235
>who do you think is training your local models?
Where do you think the training data comes from? It is just scraped from the internet. Anthropic stole the data just in the same way anyone else can. There is absolutely nothing Anthropic can do that others can't. No value.
>but but training needs stronger hardware
Cool, maybe it needs a mid-sized business to spend a few million to buy it? And then? Or it needs China? Who cares! If everybody with money to spare can do it, i don't see a purpose in Anthropic.

Amazone warehouses do not need Anthropic. Anthropic never even saw the inside of those. They know as much about those automated warehouses as we do: They watched some media reports.

I do not see a future in AI-as-a-service.
>>
>>62948384
No, the overwhelming supermajority of training data these days is generated by massive-scale RL rollouts the AI labs perform.
>>
File: melty.png (1017 KB, 1024x1024)
1017 KB PNG
>>62948384
>Where do you think the training data comes from? It is just scraped from the internet.
Wow anon. I didn't know anyone could just "scrape" the internet and get a world class intelligence model just like that!

But then you completely failed to tell me when is the next "local" model coming out then?
Hmmm. It's like you have no brains at all on this subject.
>>
>>62937993
I'm a researcher. In biology and chemistry, the models are still trash. Amazing at coding, I don't even touch code any more, amazing at administrative formalities, but I've barely noticed any improvement in basic scientific reasoning. Which is good, I might have a job better than pipette monkey for a few more years.
>>
File: 1794857902181095.jpg (35 KB, 249x249)
35 KB JPG
Im not an anthroic baggie but the premise is brilliant. Give everyone dirt cheap tokens until they forget how to do anything at their jobs and then jack the price up 500%. There will be growing pains along the way but I will ape when the time is right.
>>
>>62948706
Boomers will go back to putting grammatically incorrect sentence fragments into their emails and nothing will be different from 2023.
>>
>>62948663
It's true. You just scrape it all up.
>>
>>62948708
The boomers are retired and gen x / millenials are not going to learn how to do their jobs again.
>>
File: ezgif-7894a3995467c53b.jpg (237 KB, 1400x700)
237 KB JPG
>>62948691
>Which is good, I might have a job better than pipette monkey for a few more years.
Everytime I hear someone say that, a new AI model comes out and makes them obsolete.
2026 was the year Mathematicians got slaughtered.
2027 will probably see the first medical droids perform surgery.

God, I love the future.
>>
>>62948736
One step at a time
AI will eliminate a lot of the wagie drone workforce (McDonald's, starbucks) by the end of the decade. It will take a few years longer to make its way to specialist fields
>>
>>62948736
Anything requiring robotics is not going anywhere fast. For whatever reason, decent motor skills are the last thing they'll manage to eliminate. Turns out throwing a ball made me more of a human than anything else.
>>
>>62946136
> Splurting out a 6 gorillion page "proof" is very different from actually making a working algorithm that can defeat real encryption.
this. Same problem with their Navier Stokes result.
>I’ve proved that all problems can be solved in polynomial time!
>cool, how do you solve this problem in polynomial time?
>oh I have no fucking clue. But I can assure you, it’s technically doable.
A proof is meaningless if it doesn’t enhance human understanding.
>>
File: chips.png (536 KB, 1106x1176)
536 KB PNG
>>62948748
>>62948742
RSI will speed up research and training.
It's already being used to design and make better computer chips.
>>
I see a lot of people ITT talking about what AI will be “capable of” but that doesn’t matter if it isn’t cost effective. Yeah you can scale a neural network to 30 trillion parameters and feed it 6x the internet worth of data and it can probably do lots of amazing shit, but why the fuck would you do that?

Redstone is Turing complete, I could program the entirety of Microsoft Office within a Minecraft world. Does that mean my Minecraft server instance is poised to replace Microsoft Office in businesses worldwide?
>>
>>62948755
>A proof is meaningless if it doesn’t enhance human understanding.
There's nothing special about humans.

When the first cavemen used fire, do you think they understood the math or science behind it?
Yet by your logic, they should have stopped playing with it.
>>
>>62948785
So your argument is, we’ll just ask the AI how to solve a problem in polynomial time? But the AI is no different from a human, just proving something is doable doesn’t tell it shit about how to do it.



[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.