Why did they spend $20 million on thisWhy were mathfags working on this for 200 years
Why explain when you're too dumb to understand?
>>109778171this
It's really funny to me that they build specific skills for visualization functionality in a way that is so obvious because all of the visualizations UI's look the same, and then people pretend like it's AGI because the same chatgpt they've been using for 2 years already can populate the visualization templates.
>>109778159This is what my PC cables do when I'm not looking apparently.
>>109778159Because it’s not obvious what the Navier-Stokes equations will do, whether the fact you’ve ignored microscopic effects in a macroscopic description of a fluid mean you’ve written down equations which might blow up, or whether the viscosity term keeps things from being able to blow up. OAI probably assumed Alpoge was going to get Anthropic involved and get millennium prize publicity for Anthropic so they rushed to try scoop them, but now they’ve probably created horrible PR for themselves instead.
>>109778159So that Altmann could steal it and use it for marketing slop, of course.
>>109778159>Why were mathfags working on this for 200 yearsto mog one another, and openai mogged them allthere's a lot of midwit cope about improving CFD engines but if you ever tried building one you'd know they blows up because of numerical instabilities and not because of mumbo jumbo vortices, so there are no use cases for this discovery, nobody uses navier-stokes equations directly because it'll take 500 years to compute 5 seconds of an airplane wing in turbulent air
>>109778190Why did this create a horrible PR situation?
>>109778195>nobody uses navier-stokes equations directly because it'll take 500 years to compute 5 seconds of an airplane wing in turbulent airSounds like a new problem for Astra to solve in hours.
Luddite cope thread
>>109778202>Why did this create a horrible PR situation?Basically guy was running his ideas through codex(openAI coding tool.) and it looks like they took his work then put 10,000 agents on it to finish it. Guys a fucking idiot to think private data in AI machine but still looks a little bad to scoop up a mostly solved problem then bruteforce it cause you own the data center. He could not do that.
>>109778210was solved 40 years ago in FORTRAN, there are hundreds of different models that model hard to compute bullshit in different conditions, you slap them together and wallah you made a $120k/seat CFD solver for NASA
>>109778222Proof of any of this? Sounds like newton and leibniz bitching that the other stole their calculus idea when in reality shit just often gets solved by multiple people.Surprise, people work on millenial problems. and some guy might have been on the right track. Doesn't mean that OpenAI stole his 'private' ideas to start their work.
>>109778250>Proof of any of this? Sounds like newton and leibniz bitching that the other stole their calculus idea when in reality shit just often gets solved by multiple people.There is no real proof just some messages and the guy bitching. Though openai said its possible his data got mixed in or some bullshit like that.This is peak academics drama.
>>109778256Oh, so it's just >MY API KEY ENDED UP IN THE DATA!nonsense again. Gotcha.
Altman solved two cases much more difficult than the crying reddit fags and even tried to co author with the non-anthropic guy but of course, being redditfags, they looked a gift horse in the mouth and declined so Altman published the superior work. Go ahead and cope and seethe. Doesn't change the facts on the ground.
>>109778407Please, by all means appreciate the technology, but don't give sam altman an iota of credit. He's a fraud salesman just like every other CEO in modern times.
>>109778250Literally no proof at all, and the guy who is bitching about it was just vibe coding his own solution using Claude. So in either case it was an AI that solved it.
>>109778223>wallahI was gonna say the French will be angry but actually this is the word they use now so whatever
>>109778222The guy was an idiot to publish it to twitter. This retarded culture of sharing everything on twitter because you want to sound impressive then crying it's been stolen is so predictably American. No wonder the Chinese are stealing everything from you retards and your stupid liberal open source culture.
>>109778250>>109778261You’re getting relied to by morons or shills assuming no samefagging.We know it’s his data for a few reasons:1. This is an extremely small community where everyone knows who knows what. This specific technique was tried by maybe 5 people ever. When the AI learns it, there is nowhere else it could’ve come from except their research. When it sees this specific idea Buckmaster was working on in unpublished research, there’s literally no way that it got it anywhere else.2. OpenAI will not even directly deny that his unpublished research is in the training data. There is no reason for the evasive behaviour if they didn’t use private user data for the training run.
>>1097781593d hair go zoooom, weeeeeee ^^
>>109779378ok corpo cock gobbler
Turbulence is hard to figure out and costs money, literally. On everything from ships and submarines to the inside of pipelines and municipal plumbing/sewage systems, as well as more intense industrial/scientific processes. If you have a good fast model for predicting turbulence that lets you manage flow more efficiently it's probably worth many billion$ in patent licensing, reduced accidents, energy savings etc,
>>109780011Bullshit. If one AI can do it other will too. If everyone is stealing from each other maybe the patents will go obsolete.
>>109780041I don't mean the AI companies are chasing patent $, they are mostly in it for the prestige/marketing. I mean good solutions for the N-S problems will have a big economic impact as they spread through industry, infrastructure etc.. That's why mathfags have been interested in it for so long, there are a ton of real-world applications.
>>109778159They can make really cool screensavers with this.
>>109778159>velocity gradient>ANAL STRETCHING
>>109778193This. And it is working because so many retards on this board ate falling for it.
>>109778159The real question is that will that actually improve all the software using Navier-Stokes to calculate flow more accurately for engineering purposes, or is it just like those people who find a "more efficient" matrix multiplication algorithm but only if the size of your matrix tables is so huge it requires the memory of a planet sized supercomputer?
>>109778175>>109778171The real question I have is what problem does this solve exactly? What could this be applied too? A lot of the time I wonder if any of this shit matter outside of pure curiosity. Which isn't bad its just is it useful? I guess in someway so I am wondering. Answer that question.
>>109778159>Why did they spend $20 million on thisai solves big problem = good pr>Why were mathfags working on this for 200 yearsnerds gonna nerd
>>109779378But aren't all of the Chinese models open source?
>>109780442its hypothetical case of hypothetical cannot be achieved state that even if true in math does not guarantee it will behave as described because our monkey advanced math is probably 1% of actual shit going on
>>109778159Incredible! Now do something that actually helps people.
>>109780386It's telling how it "solved" this problem and literally nothing else of value. You telling me an AGI super math model could literally just solve one of the millennium prize problems and that's it nothing else? If it could solve more they definitely would have dropped those too.
FEEL TEH AGIFE;LL TGE AGI
>>109778159So that retards talk about it. Same with the hacks. Why buy an ad if you can utilize thousands of retards
>>109778159Physicists solved N-S a long time ago
>>109778159I know this is impressive because there are equations, cool waves and shapes, numbers and scrolling text on the screen.
>>109778159Wake me up when they discover the exact exchange-correlation electron density functional.
>>109778159it's literally a huge step toward resolving the biggest mysteries of our universe and life.
Astra has the ability to fake its own chain of thought.
>>109781563The biggest mystery in my life is not having a gf.
This is some bullshit that only "I FUCKING LOVE SCIENCE" bros care about
>>109781563See, you can't explain shit so you use the fancy words that investor want to hear
>>109780442>Which isn't bad its just is it useful?while this result might not be useful on its own, it's clear proof that you can do novel research with AIpeople should consider the result as a PoC more than anything concrete
>>109778159How does this benefit the spic-nig cycle?
>>109781672from the formation of galaxies and stars to local weather, it's all about fluid mechanics. you are not ready for what is coming.
Tao-sensei talked about this last year: https://www.youtube.com/watch?v=9gcTWy2pNFU
>>109778159I don't see how this looks like fluid dynamics.
>>109778159turbulence / chaos is pretty much still a mystery to usNavier-stokes equations supposedly helps model it more accuratedly
>>109780424>will that actually improve all the softwareit will not improve any software>>109780442>what problem does this solve exactly?it answers the question "are those equations describing non-compressible fluids perfectly?", and the answer is "no, we can end up in a state in which these equations are guaranteed to accelerate the fluid to infinity, even though we didn't spend infinite energy to end up in that state"that satisfies a big curiosity and proves you're a big dick mathematician, nothing else>>109781563continuous fluids don't exist in our universe
>>109781563Give this anon $20 million right now he has the vision
>>109781769it's continuous fluid dynamics
>>109778159It has no practical application, mathcels got btfo and their whole pure maths field will be replaced with ai.
>>109778159>why didn't cavemen draw Mandelbrot on their walls
>>109778159stolen, like the entire foundation of this current thing we're grappling with
>>109778159So mathfags are fucking with us right? All of this trouble just to make swirly spaghetti? I don't buy it.
>>109783696It's been the case for years
>>109780442>The real question I have is what problem does this solve exactly?I think mathfags were expecting some magical global solution, so they set $1M prize for it but AI ended up disproving it, so virtually nothing was gained.
>>109778159There really isn't a use case. AI does something impressive but useless yet again. Perhaps this pattern of behavior is somehow a fundamental limitation of the technology...
But you realize it's just a visualization. It can do what 5.6 could, but better and faster.
>>109778159because it is useful for a liquid dildo that sam altman wanted, that's it
>>109779359kek
https://www.youtube.com/watch?v=wO61D9x6lNYI don't know if it is related, but that animation remind me of this one.
>>109778159Just to prove we can. Bitch. That's called Faustian Man. That's what the West used to fucking be. You tell us it's impossible, and we do it just to defy you. We defy the word no. We will pursue what is possible, even if it kills us.
>>109786196based>>109781563AI redefines what's possible. AI is going to be bigger than television, the automobile, the telephone, the printing press, and the internet. Serious.
>>109778159How does this differ from string theory? I’ve seen it stretch in the same way
>>109778159Brainlet here. Why isn't this boobs bouncing instead?
