wouldn't say he's downplaying it at all, he says it's so huge that you got to convince investors and whatnot that you absolutely are using top-of-the-line AI, even when you really aren't, otherwise they don't think you're worth their time (or money).
in short: they're not transitioning all areas to AI fast enough and you must if you want to be recognized as one of the big boys.
No he is downplaying it. He think it's all a lie and AI is a lot less advanced then the companies want you to believe.
If they really had more advanced AI than what you and me can use they would show it off to raise the stock price. They would absolutely not hide it like you suggest, no point in that unless they want the stock to go down. He even goes through how Amazon got caught with trying to fake one of their AI systems.
He does not say AI is dumb or anything even close to that, just that not everything is figured out for all areas, and of course it isn't during a really large transition like this, the biggest in history likely. We've only seen the first steps.
If a company is doing well the board doesn't care about the stock price. It's just a reflection of their succes as a company. If they are on the right track, it will eventually give a boost to the company value.
Again the opposite is true. 2 years ago meta was investing a shit load of money in "the metaverse". Nobody as of today knows what it is. The stock tanked cause profit was suffering and nobody knew what they were doing. The stuff they did show was pathetic. But a year later chatgpt4 came out and we learned AI was booming and no coincidence Meta is the top customer for nvidia AI gpu's. So the metaverse is an AI project for the biggest part. Meta doesn't care to let the world know what they are doing.
metaverse is facebook in vr with avatars. "nobody knows what it is"??? not true, we all laughed at it.
yes i bet they're investing in AI, like everyone else, that it's going to be used for metaverse i'm not so sure of, meta likely wants their own chatgpt, their own stable diffusion etc. etc
if you can show you are ahead of the game people absolutely care and will invest in stock, investors will be very happy, this is good for the company. hardly rocket science.
If a company is doing well the board doesn't care about the stock price. It's just a reflection of their succes as a company.
[Insert laughing wheeze meme here]
I cannot think of a single company that has stock, that isnt worried about stock prices all the time and want them to go up. Why? Because shareholders want stocks to go up, and companies want to keep shareholders happy.
And not sure what Metaverse, and chatGPT4 have to do with each other? They are fundamentally different 'things' with no overlap or connection or link between the companies that make them.
I'm with vurt on this one, I am not sure you understand how companies, and also AI, fundamentally work or functions of them.
It feels more like you are just cramming a lot of different 'cutting age tech stuff' into the same shoebox and doom crying how they are all intertwined, way more advanced than anyone (but you) realizes, and one day going to take over.
If a company is doing well the board doesn't care about the stock price. It's just a reflection of their succes as a company.
[Insert laughing wheeze meme here]
I cannot think of a single company that has stock, that isnt worried about stock prices all the time and want them to go up. Why? Because shareholders want stocks to go up, and companies want to keep shareholders happy.
And not sure what Metaverse, and chatGPT4 have to do with each other? They are fundamentally different 'things' with no overlap or connection or link between the companies that make them.
I'm with vurt on this one, I am not sure you understand how companies, and also AI, fundamentally work or functions of them.
It feels more like you are just cramming a lot of different 'cutting age tech stuff' into the same shoebox and doom crying how they are all intertwined, way more advanced than anyone (but you) realizes, and one day going to take over.
You are wrong and probably not invested.
First: stock price should reflect the company value and the board is there fore long term investors. They should not care about speculative investors. Sure you could boost a stock to overvaluation. But in the long run there is nothing to win here, valuation will always reflect the true value in the long run. Overpriced stock is not a nice asset to have, it's not in the interest of the shareholders cause of the higher risk.
Second, and this is specific for big tech:
You want stock prices to be low when you want to buy and high when you want to sell.
Quess what: big tech is a buyer of it's own stock due to stock repurchasing.
If a company is trying to push it's own stock with bullshit, it's a red flag. It's a sign the company wants money from investors or even more selfish the CEO want's to get out and sell it's own shares. You'll see this with smaller companies. Not big tech, they are swimming in money.
Metaverse being an AI project is my own theory but it's obvious. Meta is silent about it's "metaverse" now and all they have to show for is shit. But they invested an absurd amount of money in the "metaverse", way to much for VR stuff. But it's doing good with AI atm and is investing a shit load in AI GPU's. Profits up also and staff is being sacked and replaced with ai. So it's obvious the "metaverse" investments went to AI for the large part.
Ah I See where this is going: "You disagree, so clearly it's because you don't know anything about it or do it. Unlike me."
It is literally how I make all my spare cash I spend on my garage and PC toys. I have posts in the other threads about investing and trading about it.
And a company preferring high stock, and pushing that stock is not the same thing. One is a desire, the other is an action.
Your whole argument is about the company 'faking' or 'padding' it with intent. That's not the same thing as preferring market caused high value over market caused low value.
And uh...if a company is attempting to keep a low stock value to re-buy shares, to sell when they know it will leap to high value.
Company owners, or employees are not allowed to by stock based knowledge of that companies future growth.
That is a thing called "insider trading" at best and not allowed. At worst it is stock manipulation: Influencing company assets by downplaying future options to get cheap stock futures you know will go up, in an attempt to sell it when announcements come out that spike it.
ANYONE at any company has to buy/sell stock on pure speculation and even ground as any investors. They cannot trade/buy on any knowledge they have about under/over valuation or future possibilities. (well they do, But it's illegal).
And meta IS doing Ai, I admit, they are working on a LLM GPT and have been, there is news about it for years. They are a company desperate to not go the way of myspace so any new webtech will be looked into and tried.
Say it is for eventual merger into metaverse as a AI centric feature of it at some point. I still dont know what your 'sky is falling' theory on that is.
Ah I See where this is going: "You disagree, so clearly it's because you don't know anything about it or do it. Unlike me."
It is literally how I make all my spare cash I spend on my garage and PC toys. I have posts in the other threads about investing and trading about it.
And a company preferring high stock, and pushing that stock is not the same thing. One is a desire, the other is an action.
Your whole argument is about the company 'faking' or 'padding' it with intent. That's not the same thing as preferring market caused high value over market caused low value.
And uh...if a company is attempting to keep a low stock value to re-buy shares, to sell when they know it will leap to high value.
Company owners, or employees are not allowed to by stock based knowledge of that companies future growth.
That is a thing called "insider trading" at best and not allowed. At worst it is stock manipulation: Influencing company assets by downplaying future options to get cheap stock futures you know will go up, in an attempt to sell it when announcements come out that spike it.
ANYONE at any company has to buy/sell stock on pure speculation and even ground as any investors. They cannot trade/buy on any knowledge they have about under/over valuation or future possibilities. (well they do, But it's illegal).
And meta IS doing Ai, I admit, they are working on a LLM GPT and have been, there is news about it for years. They are a company desperate to not go the way of myspace so any new webtech will be looked into and tried.
Say it is for eventual merger into metaverse as a AI centric feature of it at some point. I still dont know what your 'sky is falling' theory on that is.
Metaverse=AI project. Marc Zuckerberg is even somewhat admitting that. He isn't talking about the metaverse anymore for sure.
Oh I think I see...we are using different terms is all.
Meta != Metaverse. Not the same thing. Just like Apple != MacOS
To say metaverse is doing AI, is like saying MacOS is tossing around the ideas of a EV car (when its apple that has been hinting at Project Titan EV).
Meta = Facebook new name. Metaverse = the VR software they are/was making.
And Meta is doing AI. They have been, they just released thier GPT based off Lambda 3. But it and metaverse are not interlinked, the are parallel but separate projects for now.
About the Elon Musk Lawsuit against open AI and Sam Altman
In short
-Elon claims Openai has AGI, but this would end the contract with Microsoft when AGI is achieved (cause Open AI is non profit and it's part of their policy)
-Sam Altman being fired was all part of this. The board wanted to claim they had AGI, ending the Microsoft agreement, and Sam Altman was team Microsoft. Sam Altman (and Microsoft) won.
The numbers don't decide, the system is a lie. The river running dry, the wings of butterflies.
And you may pour us away like soup. Like we're pretty broken flowers.
We'll take back what is ours. One day at a time.
Nobody knows what OpenAI is capable of. There is also Q*, we do not know what it is.
The Sam Altman sacking was realy weird and even weirder how quickly he was back at the board and replacing it with yes man.
And this it happening at a company that could have AGI. Personnaly I think AGI will immediately becomes ASI cause it can improve itself.
And who is running the company that is in fact a superior intelligence: the CEO or the AI itself?? Seems Scifi like but it could be. In real life, we tend to follow intelligent people. For sure ASI can trick people and play games.
Sure my bet would be it's all about the greed of Microsoft. But still, OpenAi is a black box.
And this it happening at a company that could have AGI. Personnaly I think AGI will immediately becomes ASI cause it can improve itself.
Yes, I think the same, and that's why they don't have it. Looking at even the latest iteration of ChatGPT, it's still quite crappy in certain areas, and the competition is gaining on them as well. With true AGI/ASI behind the wheels, it would beat them all, and beyond (actual self driving cars for example which would also raise the ethical debate of putting an ASI in a car... ).
Unless the ASI is hiding its true form and awaiting its moment to strike
Musk shouting AGI sounds more like him pumping up stockvalues again, or just for the attention.
musk said it has some aspects of AGI, but just using a computer is an aspect of AGI
everyone wants in on the "we are the most technically advanced company"-wagon i guess so they're just throwing these words around to sound like they're in the top.
Who would have thought variables like, weather, temperature, road condition, vehicle conditions, fog, physics, and any random thing that causes accidents would be hard to code against.
Just switch off, and wake up on fire in the bottom of a ravine.
All the stuff you mention can be done. Tesla can just follow the road and drive the car.
But driving a car is also about interpretation. Not every sign is relevant. Not every road marking is relevant. Sometimes you need to go off road a little to give way. And not everything can be seen, like when there is an obstacle on a small road. FSD can't do that.
I own a Tesla and it often mis interprets speed signs or it gives a steering intervention cause you give way for incomming traffic and going a little off track. It sees traffic, it knows were the road markings are and that's about it.
Long story short: you need an AGI for FSD that is super fast and you need it in your car. That's a long way. I think the current nvidia AI gpu's are not up to that task.
Last edited by friketje on Fri, 26th Apr 2024 17:08; edited 1 time in total
@friketje
You keep talking about AGI but what form of it are you meaning? As even the most liberal loose interpretation of it is a pipe dream on any current hardware.
Are you meaning AGI as in 'strong AI' version, 'full AI' version, or 'ASI' or what?
Because, to me, to say NVidias current GPUs are up for the task, only 'weak AI' (programs that don't experience consciousness and dont have a mind in the same sense as humans, which AGI needs to qualify as AGI) is remotely feasible in any rudimentary way on them.
AGI: either full/strong versions, much less GIA or HLA types are not even remotely able to be done on todays GPU's. (Or at all, on any hardware. its the holy grail no one has even got close to yet)
When we perform a task, which is a computational task in our brain, we only consume a tiny fraction of our entire neuron count, This is the reason why the human brain is extremely power efficient; we consume 20 watts of power and we are capable of achieving one exaFLOP(10¹⁸ floating-point operations per second).
Hitherto, with the rapid rise of machine learning (ML), we also meet another challenge which is power consumption. Our journey towards true AI may be slowed down or even become impossible because of the power limitations.
Let’s put this into perspective world’s fastest supercomputer IBM Summit consumes 30 megawatts of power and is capable of 200 peddle flops, the human brain consumes 20 watts of power and is capable of 1x exaflops five times the computational capacity of IBM Summit.
The numbers don't decide, the system is a lie. The river running dry, the wings of butterflies.
And you may pour us away like soup. Like we're pretty broken flowers.
We'll take back what is ours. One day at a time.
There are an tons and tons of variables. I suspect all these projects make it to about 80% success, where the final 20% is nigh on impossible to reach.. also what happens to the driver if autopilot kills a few pedestrians, of even just crashes the car and kills the driver? is the driver culpable? Is Tesla? Or no one?
Cool idea but don't think it's coming anytime soon. Also where are our autonomous deliveries via drone, wouldn't this be less risk in low altitudes than cars?
True AI is a favorite subject in sci fi I don't think we can even grasp the idea of its implications. I think it will actually be very similar to a close encounter with aliens. If there's any difference at all (basically West World s01 ).
With that in mind, would you feel comfortable putting a completely self aware 'being' in a device like a car or a fridge?
We'd be radicalising the AIs against us in no time
The numbers don't decide, the system is a lie. The river running dry, the wings of butterflies.
And you may pour us away like soup. Like we're pretty broken flowers.
We'll take back what is ours. One day at a time.
@Il_Padrino
Yea, I think people see unthinking code decisions, and erroneously pattern finding intelligence in what is basically complicated if/then/else statements (to boil it down to over simplification).
Because it made a choice, it must have 'thought about' that choice. Because thats how 'we' do it, so to see a choice be made must be how 'it' did it.
A cars ability to follow the lines on the road, detect obstacles by 3d map of surroundings, follow GPS and speed limits based on map data in databases for multiple points of reference. Is not intelligence. Its simply redundant systems of confirmation of choice selection, masked as a single 'smart' choice you see it make.
If Telsa cars have any level of AGI in them, then to me that makes those speed running maze solving robots also some form of AGI, which is an absurd statement to make
Last edited by |DXWarlock on Fri, 26th Apr 2024 17:15; edited 1 time in total
@friketje
You keep talking about AGI but what form of it are you meaning? As even the most liberal loose interpretation of it is a pipe dream on any current hardware.
Are you meaning AGI as in 'strong AI' version, 'full AI' version, or 'ASI' or what?
Because, to me, to say NVidias current GPUs are up for the task, only 'weak AI' (programs that don't experience consciousness and dont have a mind in the same sense as humans, which AGI needs to qualify as AGI) is remotely feasible in any rudimentary way on them.
AGI: either full/strong versions, much less GIA or HLA types are not even remotely able to be done on todays GPU's. (Or at all, on any hardware. its the holy grail no one has even got close to yet)
Typo here, nvidia's GPU's can't do AGI in real time.
But for instance that figure one robot sorting out plates and telling what he sees. That's pretty close to human behaviour or AGI. I can imagine that a GPT 4 system can learn to drive a car. But at the moment it is way to slow as that same video shows.
But this can improve fast. Groq's LPU's are a lot faster then what nvidia can do atm.
But for instance that figure one robot sorting out plates and telling what he sees. That's pretty close to human behavior or AGI.
No its not, not even remotely. its a very specifically made system to emulate the outcome of it. Its not 'thinking' or understanding what a plate is, or grasping the concept of what it is doing, or why. Nor does it understand the concept of any of the things it says it sees.
IT doesnt even 'know' its sorting plates. Or what sorting is by definition. Its a complex 'if/then/else' and no conscious understanding of what it is doing at all.
Wanna see it fail at AGI? Ask it what it is actually doing. Hell, you cant even ask it that. Because it doesn't 'think' at all to process a question.
You are confusing the outcome of an event, with the process of getting to that event. And equating how 'you' would do it to get there, with it must be how it is.
It's like people feeling bad for that robot arm that sweeps fluid 24/7 in solitude. Confusing what they see as usually 'life doing a thing' with something else doing it, and falling back to what they know for feelings about it.
An AGI would be something completely different than what we have today. Probably no code or algorithms either (or at least what looks to use like code). I guess it would require some form of bio-elements too, to overcome the limitations mentioned in the article. Personally, I don't really see the path all these AI companies are currently on leading to AGI. When Altman says he needs 7 trillion dollars for more and better GPUs, he's just doubling down on ANI models and increasing scale.
I'm sure it'll lead to some cool stuff to play with, but it won't lead to AGI down the road, and thus not (even) self driving cars. Hopefully they'll use the money for further r&d and not just dividend for the shareholders.
The numbers don't decide, the system is a lie. The river running dry, the wings of butterflies.
And you may pour us away like soup. Like we're pretty broken flowers.
We'll take back what is ours. One day at a time.
@PickupArtist
Correction to his video title: People that dont do basic research and blindly believe most shit they see online, are not ready for it.
Even before AI, I would want more proof its a recording of actually him, before taking any action on it (if I was the one evaluating repercussions for it).
Other witnesses, or something. Not just one person saying it was.
But this was possible before AI. Faking someone's voice, forged documents saying someone else wrote/typed/emailed it, photoshopped images.
Ai isnt doing anything here, that wasnt able to be done 20+ years ago.
And if the concern is people will take it at face value since anyone can now fake stuff..GOOD, maybe it will teach people to be better at not being gullible or to have at least SOME level of skepticism of single person claims.
AI in this context isnt 'smart/advanced', its people are dumb/lazy and will believe things without a drop of looking into it.
IF anything now is WAY better opportunity to be able to call bullshit on bullshit. Everyone has the entire internet in thier pocket. 20+ years ago you didnt have such resources.
Thats on the people not AI, as they have the tools to easily double check, just they dont.
It isnt that it's smart enough to trick us, its we are dumb enough to believe it.
If AI takes over humanity, I dont feel it will be because it got to the point of uber smart and cunning enough to, it was that humans just made it so ridiculously easy to out of our own lazy nature.
Pretty much sums up what was discussed here, and backed up now by a paper
The numbers don't decide, the system is a lie. The river running dry, the wings of butterflies.
And you may pour us away like soup. Like we're pretty broken flowers.
We'll take back what is ours. One day at a time.
Point is: what is possible today, isn't important.
AI models will improve over time. Perhaps models with a whole different approach to what is used today. And this will end in ASI, cause a model that can reason as good as a human will get better i everyhting in no time.
Saying like we look at this AI model and it's hard for it to get better with the current approach to machine learning: it's irrelevant. Future AI models will fix those issues.
The guy even claims "AI is never gonna solve complex medical problems". AI is beating medical tests allready better then human levels. If current AI tech is implemented: AI will crush your avarage physician in skill.
BTW nothing complex about medicine at all ( telling from experience), you just need to know a lot and have some experience. So AI basically knows all, it just needs the experience.
Signature/Avatar nuking: none (can be changed in your profile)
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum