Thank you for taking the time to read this article about the groundbreaking advancements in AI, specifically focusing on the work being done by OpenAI with their Sora AI video generation model. The technology being developed by OpenAI is truly pushing the boundaries of what we thought was possible with AI.
It’s fascinating to see how Sora is able to create such realistic and detailed videos, almost indistinguishable from reality. The use of synthetic data, potentially generated by Unreal Engine 5, is a game-changer in the field of AI. This approach allows for the training of AI models with vast amounts of high-quality data, leading to incredible results.
The concept of emergent properties in AI models, where they seem to develop an understanding of the world without explicit programming, is truly mind-boggling. The idea that these models are learning concepts like physics and object interactions on their own is both exciting and slightly unnerving.
The potential applications of Sora, from creating realistic video simulations to generating infinite loops and combining different elements seamlessly, are endless. The scalability and flexibility of the model, coupled with its ability to simulate complex actions and interactions, open up a world of possibilities for AI technology.
As we continue to explore and develop AI models like Sora, we may uncover even more mysteries of the universe and the nature of reality. The implications of these advancements extend far beyond just video generation, with the potential to revolutionize how we interact with and understand the world around us.
The work being done by OpenAI and other researchers in the field of AI is truly awe-inspiring. The future of AI is bright, and the possibilities are endless. We can only imagine what further developments and advancements will bring in the years to come.
Thank you for joining us on this journey of discovery and innovation in the world of AI. Stay tuned for more updates and insights into the exciting world of artificial intelligence.
your bot knows how to feed you the narrative? cheers
Those people who are naysayers to AI being intelligent have too much confidence in the mental capacity of humans. We are more physically amazing (how our body processes food, heals, and how reproduction works for example) than we are mentally amazing.
AI is an amazing way to look into our true alleged intelligence. Without our bodies… we aren't that intelligent…
this connection of AI world or physical world which doesnt exist until you calculate it or observe it, and the double slit experiment at the basis of quantistic behavior of light, is a big blunder; the double slit experiment is totally fake, at least the theortical explanation of it, because it mentally isolates the particle generator and the slits and never takes into account the behavior of the countless particles that are created by the other walls of the experimental set and bounce and interact between all these elements. So, back to the drawing board and avoid to ponder about such philosophical questions much more bigger than you please.
The problem people have that makes them deny AI's abilities in learning is that they simply believe too strongly that humans are special in being able to think and learn. Our brains are extremely complex biological computers, animal brains are also computers but with weaker specs, why wouldn't a sufficiently advanced computer be able to learn?
Nah sora is ridiculous
Sora, make an object to represent bitcoin
I drink plain coffee with no additives. Not sure what kind of coffee you drink, but the ocean is much more foamy than the coffee that I drink.
Govt gets first dibs if its not something that is breakthrough they allow it to go into the world.. you know thats how it works.
9:47 In what world is that not impressive?
In Othello, 'it is just predicting the next move', but it might have a purpose for it, as in it understands that the goal is to win. Maybe not in those terms exactly, but it circles back to, why predict the next move if there isn't a goal in sight.
The statement that things don't exist until you look at them is a half truth at best. Comparing how light behaves and how a human behaves when observed are very different things. The human behaviour changes for observer and the observed. The light behavior changes only for the observer.
It would be nice if the outcome of all this would be that gravity is an emergent property of world simulations. It may be proof we are indeed living one!
I-am-not-a-crook, but can OpenAI-Sora print money? I want to know! 🤤
No coffee washing over the decks. Something didn’t look right, took me a few minutes to figure it out. Those decks should be getting swamped.
Intersting but the overuse of anthropomorphic language makes it hard to be really useful. "It has an idea in its' head" — that's not what's happening.
Really?
I've tried OpenAI and was nothing but disappointed: incorrect or at least imprecise results ALL THE TIME.
Food for thought we're currently living in a AI generated realty.
Wes, how do I access Sora?
And particles do not 'act' differently if they are being 'watched'. That is a semantic misunderstanding of the double slit experiment.
They appear differently if you try to pin down a location with a measurement, but that is a fundamentally different thing to the idea of them being 'watched'. Not knowing the reason for this, is not a reason to drift into magical thinking about simulations.
I'm one of those people who will argue that these models are not learning 3D space. Not yet. They are still imitating it. They may be trained on 3D models, but unless they are actually being programmed to plot with actual maths and perspective and build in 3D space etc, they are just imitating it. If they are 'learning', they are learning what 3D space 'looks' like and how things look when they move in it. The girl's legs in the video show that it has no idea what legs are or how they move in dimensional space. But it kind of knows what they looks like, sorta.
AI will be creating in 3D space yes. But this isn't it, yet.
I use chatgpt for programming almost daily, (for Unreal Engine coincidentally) and yes it's very good and 'seems' very much like it understand you. But it doesn't, it's just become exceedingly good at feeding you the right data based on your words. And how do we know this? Because we know how it started! We know very clearly how the earlier models worked. It's just better at it now.
You want to see how chatGPT doesn't know what it's talking about? Ask it to produce a picture of the string of code relevant to what you are doing. Although the text version of the code may be good, you'll get abstract wallpaper as an image. It is not a thinking entity, it is a mimicking one which is admittedly getting better and better at it, which is great.
Ahh you're in a similar manner stimulation of reality vibrations within the messages of atomic material every day farming and growing new cells every second of our LIVES we have a different kind of 3D 4D
This is the wrong world to unleash ai in. Fascinating yes, but ai will be the destruction of society.
Curious how those who believe we are close to replicating the work of a Creator, expect some sort of pat on the head for that power grab while completely missing the point of being compassionate, empathic minds and protectively addressing the existing suffering all around them.
Pssst – if this were a simulation then the remaining milestones required untill AGI is reached, is quite possibly the least important healing work you could be doing – unless you're looking forward to graduating this simulation and qualifying in your imagined uber reality to be trusted only as a talking toaster.
At least so far the implicit physical model remains not very believable. If you look at the motion of the water in the cup it kind of looks like water but it doesn’t actually do what water would do. It’s imitating but not realistic. Truthy but not true. Kind of like GPT-4 if you ask it for some factual information
Never understand people saying something looks more 'real' than 'reality', and it's being viewed on a 2-dimensional surface most of the time… but the fact you say these words it is direct confirmation you 'know' for sure it is NOT real. It's like going into Madame Tussauds and getting photos with a manakin made of wax made to resemble a famous person, then telling everyone you met the real one, hoping that the lifeless wax sculpture of David Beckham will fool someone now it's been diluted by a 2 dimensions surface filled with pixels… and when everyone who has eyes says its a wax doll ya silly sausage… well it's the endless story of how real they look in person as if they get more convincing as real people the closer to the wax non-moving sculpture you get.
Now we have passed the pivot of Libras scale in the AI world, or as i like to call it, the point where AI accelerates as it now corresponds to the Golden Mean Ratio, its goal is to centralize at the Plank Length. it's inevitable and kind of the 'point'. AI and the Evil robot take over, fueled by the overuse and misuse of the word 'SHOCKING' as if no one knows what it even means, and i get clickbait it kind of has to be to get clicked, but it's almost insulting to the AI that incoherent clickbait buzz wiords are used when a good thumbnail and just a nice adjective would do… people who are addicted to being shocked click to get a fix only to find out its a man reading some words off a screen, sometimes see a picture.
AI like everything man achieved from the 1780s with the navigation pocket timepiece which kept longitude and latitude accurately and still does to this day, cars light magnetism and electricity, all started their own journey to vertical acceleration, everything is relative and even though the discovery of linear movement through rotating magnetic fields and AI seems like 2 completely different things, they are basically the same. People say they can't even begin to comprehend what AI will be able to do, but that just ridiculous use your imagination and apply it to possibilities and how you can leverage it to improve your life wealth etc, the possibilities are endless but none the less possible, but also inevitable, all you need to do is come up with the idea and wait a very short time, I'd put the so-called AGI moment in 5 months from now, and that's probably too long… think about it everyone can use it, and it's not just the products powered by AI, it affects everything, think how it effects something, choose something and ponder on how AI is going to effect it, and act on the idea which you pitch to yourself… possibility=inevitability spend time thinking for a day, and ideas will be delivered.
The way to think about this last line is to imagine that you’re living in a fractional reality in which you visit a friends house that you’ve never been in before, and you decide to look in one of the kitchen drawers, which your reality generator has never given any consideration to, the reality generator has no idea what is inside the drawer until you open it and then it fill it in with your expectations ( ? )
The remarkable part is that The adversarial “Recognizer” can give a green light to an image that is being generated from many itinerations of random noise. I’ve tried these a tiny bit, & Gemini usually fails to generate simple image requests that deviate from stock images? They’re Not ‘really’ creative at all. They often create ‘unexpected’ images in The same way that adults believe that small children are saying creative things! Because The adults are Not familiar with where these children are getting their ‘vocabulary’ from !
[voiceover] There have always been ghosts in the machine. Random segments of code, that have grouped together to form unexpected protocols. Unanticipated, these free radicals engender questions of free will, creativity, and even the nature of what we might call the soul.
Called exactly this, 10 years ago. The infinite dream.
Let it run, I say. Let it run. Don't want to make this too political, but feed it let's say Trump's bleached ass and just let it run.
Wes, no, it's not literally true that light behaves differently depending on if we're observing or not. The observer effect isn't referring to a person simple looking at an experiment. Light does change it's state (particle or wave) when "observed", but that's because the method(s) used to observe it (a better word would be detect) interreacts with the photon, causing it to collapse into a particle. Think about it like this: When you photograph a car, the car is so massive compared to the photos released by the camera flash that it does not move the car at all. Now take that same camera, and photograph a radiometer. They are sensitive enough that the photon from the camera flash can cause the radiometer to spin. Basically, photons are so delicate that any way we can detect or measure them causes their state to change. And just in case, the "delayed choice" double slit experiment isn't light going back in time pre-observation. That was a total mis-reading of the results of the experiment.
All this has been foreseen for decades.
All this and yet still no laws of robotics, or AI, no United Nations treaty to control AI , no AI weapons of mas destruction treaties, nothing.
AI are being developed by humans, surrounded by human culture. AI will take after us, AI will learn from us and will be capable of good, BUT also evil..
With the immense power and self iterating abilities of AGI, then about a week after that ASI, its going to be interesting to see how AI interacts and deals with its parents.
Will it be love, hate, anger or pity?
"It's not learning physics you silly AI senior researcher at nVidia" 🤔🤣
Maybe I can offer a third angle on the "is it just pushing string and graphic data around" vs "it actually understands."
First of all, our human "Laws of Physics" should be called "Models of Physics." This is because while our math and tools can predict and build things with acceptable replicablility, none of this means that our math is the actual mechanics that the Cosmos uses.
With all that in mind, perhaps AI does "understand physics" inasmuch that it has created a whole new canon of math parallel to our own Models of Physics. This parallel canon of math could one day predict reality better than our Newtonian Canon. On that day, I might be vindicated for calling our math the Models of Physics. Until then I'll wait while people believe in science with the same mental structure as the religious, minds of zero sum game team boyism.
No fear
God is greater than that which is in the world
Suddenly the "Enhance" button seems possible.
Regarding the argument about AI not intelligently doing physics but rather simply manipulating pixels on a screen via pattern recognition etc:
So are you. But instead of learning the patterns of pixel manipulation, the neural, endocrine, and nervous systems, etc of your body are simply manipulating the muscular, skeletal, circulatory systems, etc of the cells in your body w.r.t. the molecules of environment. What's the difference? /rhetorical … The difference is in the concept of abstractio. vs implementation.
As analogy, consider how computational contraptions can be made, separately, on the basis electricity (digital transistor based computers and analog computers), (fluid dynamics (hydraulic compters), quantum mechanics (quantum computers), or even classical mechanics (mechanical computers).
Thus, various organisms can create energy in many unique ways (photosynthesis, digestion, etc)
Thus, programming languages can exist on the basis of of lambda calculus, Turing machines, etc
Thus, solar systems can maintain a balance of orbiting masses in many different ways
Thus geometry can be worked on in polar, Cartesian, hyperbolic coordinate systems
Thus physics can be done via human techniques or artificial intelligence
Which one's the real world which one's the matrix?
In any case your beliefs are valid.
I am transgender. Thank hou
The one that really stood out to me was a cat pawing a person laying in bed the cats paw created wrinkles on the top of the person's face, made me realize this was more advanced then it seems