As long as Robotics is stuck I won’t say it is over. Once Robotics reaches perfect Dynamics we are screwed, but machine learning will have to improve a lot
Bro like is just me or in just 3 years we all of a sudden had AI Boom? Like before 2021, we didn't even have any AI source, except those in some website who use automated help and call it their own AI, but hasn't AI advanced so much in just some years?
a lot of the shorts uploaded are just AI generated garbage already, you can recognise it with the tiktok AI voice and usualyl the minecraft parkour background, there's full automation scripts already to autopilot channels
@@ahhhh226 oh yeah, but the rate of AI progress is definitely gonna change that sonner than we'll expect...a good measure is to estimate the percentage of people who can't distinguish between the 2..currently, It's the boomers
I dont remember where i read it but its expected that by 2026 around 90% of content on the web will be ai generated. That feels about right and its horrifying.
I looked into this and the pricing is outlandish. There's also the snag that what's been generated may have copyright issues once content ID algorithms work out where the source footage is from.
Chaque génération est unique et personne ne pourra crier au droit d’auteur. Aucun visage connu ne sera généré même si SORA en sera parfaitement capable et ni vous ni moi ne verrons la différence.
It's an interesting thought I guess, but seeing as pictures and videos have only existed meaningfully for like 150 years, I really think it's odd to think that taking pictures and editing videos define what being human is.
omg can people stop losing their minds. If people start creating movies and games with this it will still be a work of an artist doing it. There will always be human vision and skill behind this. That would be cool actually, to be able to create a movie by yourself like writing a book.
@@themartdog You're an example of someone that's not human, but rather a robot, npc or bugman. You're not able to abstract out ideas and concepts form what someone says, you can only view it literally.
For those that didn't know, the character at 0:46 is actually the character for くそ, which is used for similar situations where you would shout "SHIT" in English. The actual character for sora is 空．
@@TheFunkoDunkothis also means sky, air in Chinese... no wonder I believe. And this pronounce different in your example, not as if it were "empty", but "unoccupied"
@@sjoerdev Bro, I'm asking AI stupid questions all the time and I'm learning new concepts faster than ever before. AI will make our jobs a lot easier. It can't easily come up with ideas though, so we're safe for the time being. I used to google stuff: Now I use AI first and use a search engine in case the answers are truly bs.
It's not the "AI" that threatens humans. It's other, powerful humans and these are going to be the ones who will utilize the new technologies in order to rule over the rest of humanity
its stupid to think AI will harm us, harming is a primitive biological trait in the view of machines. we will just get insignificant automatically by ourselves when AI is better than us, not more. a realization of ultimate meaningless and nothingness in the abundance of the universe, or merging with something more advanced.
it took me time to realize but the future of this tech is probably even more insane than that since it understands 3D space accurately presumably so it means that this tech in the future might one day be able to run completely in real time making it possible to be experienced in virtual reality googles (by also splitting the same single image into two different image angles for 3d depth effect), it's so remarkably advanced and revolutionary that I think we still don't fully grasp how powerful this is going to be in the future, who knows it might even replace game engines at some point and do magical things in real time.
@@ArbitraryZer0101 I'd love that in VR or even better MR :) my hope is that other companies (free and opensource) would also develop a similar model like Sora so maybe one day it might be even possible to run it locally
@@Wooflungpoo if Elon or someone else manages to pull of those brain implant chips the algorithms wont even work off what you liked before, it will be exactly what you want to see at any given specific moment in time the instant you think about it
Reminds me of the joke people used to make about the actors in the first Doctor Strange movie: ❎ Benedict Wong ✅ Benedict Wight (shows Benedict Cumberbatch)
After 3:20 it took my brain about 10 seconds to realize "wait, did I just watch a butterfly swim under water?" I had to go back to double check. Didn't notice anything wrong with that the first time. 😂
The advancements in AI truly have the potential to induce an existential crisis. And despite the terror of being replaced by a machine, it's hard not to be impressed by the capabilities of Sora. The future is daunting but also fascinating, what an era to be living in.
You prefer existental crisis from being forced into existence as a dressed-up monkey on a rock in space and later forced out of it after likely suffering physical and/or cognitive decline for years?
The sheer amount of computing power, bandwidth & storage needed to run these models at scale is ludicrous. These companies will be razing thousands of acres of land & water sources to build colossal data centers, just so millions of people can create digital garbage, exponentially increasing storage demands even further.
It's amazing how much resources they are using to predict human behavior and to replicate billions of iterations of yesterday's data. How much power in Gigawatts it will require to run AGI? And for such a fantasy like ASI they'll need the energy of the entire Sun.
@@oldclient And all it will take is one screwdriver to take down the entire thing and their dream of predicting the human behaviour and every thought. It's actually hilarious when you think about it.
Fr. I'm like capitalism, but increasingly it seems incompatible with technology like this without any restrictions put on it. Even 2 trillion is a massive, massive pile of future e-waste we're going to need to deal with. Thanks Sam.
"I'm sorry I cannot help with that request. Further, as per our agreement with Hollywood regarding their IP, we have terminated your OpenAI account. If you feel you have been terminated in error, well, call Sarah Connor. Hasta la vista, baby!"
@@michaelnurse9089 Just have chatgpt write a good script, then based off of that script use sora to create short clips and blend those clips together to make a movie
My employment entails graphic design, web development and video production, and my hobbies include photography, painting and game development... Just point me in the direction of the Euth-a-Booths.
Singularities are physically impossible, the only thing that they tell us is the fact that we have no idea what will happen in real world at their places and times
@@akeem2983 Quote: _The technological singularity-or simply the singularity[1]-is a hypothetical future point in time at which technological growth becomes uncontrollable and irreversible, resulting in unforeseeable consequences for human civilization.[2][3] According to the most popular version of the singularity hypothesis, I. J. Good's intelligence explosion model, an upgradable intelligent agent will eventually enter a "runaway reaction" of self-improvement cycles, each new and more intelligent generation appearing more and more rapidly, causing an "explosion" in intelligence and resulting in a powerful superintelligence that qualitatively far surpasses all human intelligence.[4]_ I don't understand how this would be impossible. It seems like a relatively plausible concept.
@@CatgirlExplise6039other than all the benefits improved coding, more efficient tech development and better medical research achieves...nope, no improvements to see here.
Deepfakes, voice manipulation, and whatever else.. in the end, your subconscious doesn't know the difference so say goodbye to trusting your device because it will be hard to know what's real anymore.
SORA - upto 1min, 1080p SORA 2 - upto 10min, support 4k, upto 10k tokens SORA 3 - upto 30min, 4k, support inpaint/outpaint, support video/audio input as prompt upto 100k tokens SORA 4 - upto 2 hours, support 8k HDR, upto 1M tokens, support audio output SORA 5 - input any tv shows/books/movies, output the version you like
@@3793loop Then Apple will respond by unleashing a 100 million token iAI, only starting at a price of $300,000 for the Budgeted version which only has 20 million tokens.
If we make text to movies, then OpenAI makes GPT5, then I fear what Disney will do with their deepfake technology and combine it with the first two things I mentioned.
I'm DCAing in Blcktken300 as well. ETH heavier DCA and ALGO. I'm taking your advice and starting Google tomorrow with a 50 dollar purchase and continuing Microsoft and Apple. VTI and VOO on another app and longterm portfolio. Here we go family!
I remember having a Commodore 64 as a kid, and at the time is was the best the home computer could get. Then the Amiga 500 came out and our world changed. We went from 16 possible colours, to 4096 possible colours, and the quality of graphics in games exploded. Then there was the churn of IBM clones over the last few decades, with ever increasing steps in performance though nothing quite as gigantic in terms of the leap from the C64 to the A500, unless we include the leap from pc speaker to Soundblaster! But this AI, and what it means, and what it encompasses at its very essence, is another milestone in human history. It's no longer a case of "imagine it - create it", we don't even need to imagine it any more. A vague string like "space man wearing knitted wool motorbike helmet" resulting in a small movie is not even in the same league as where we've been stuck. I'm 20 years away from retirement as an electrical/electronic/software engineer, and while I feel reasonably safe in being able to work until that time comes, I feel genuine fear for the upcoming generations and what this will all mean to their future. AI is the literal embodiment of bitter-sweet.
04:12: “It’s been a privilege and an honour to watch ten thousand years of human culture get devoured by robots.” This might actually be the most fitting summary of generative AI.
Utopia: About we focus all AI to just some insanely complex problems that plague the world, so that we all can live in prosperity and health and those who have inspiration will have the chance to learn the crafts and make art. Dystopia: We have a world full of war, disease, hunger, and depression ... but at least we made all artists redundant.
Or, you could turn it into a never ending TV series. After each episode concludes, you prompt your AI slave to craft a fresh installment uniquely designed to satisfy your dopamine-hit dimension
Yup, we are like the Titanic's musicians: watching everything go down but with poise and dignity. nah, who am I kidding? I'm fucking scared my paycheck is gonna get deleted now. gonna ask chatgpt tips how to survive the street and the homelessness lyfestyle.
@@mernkanthri3941 I mean all those issues exist because of capitalism, not AI. Not saying AI is all good, but it's just a symptom of a bigger problem. Fighting it doesn't change the fact that rich get richer or that we will all be eventually obsolete in a world that sees us as numbers. There's only one solution to fighting that, and it's by attacking the root cause itself
@@mernkanthri3941Thank you! It's refreshing to see someone have the same thoughts as me. Do you want to join hands with me? I think we should launch a full fledged revolution?
Oddly enough, I do digital art and business has never been better than since genAI came out. Either people tried AI, and found out it's extremely limited if you need something done to exact specifications, or they don't like the idea of AI, and support human artists to balance it out.
Lets enjoy however many years we have left where we can still say for certain that the videos we are watching aren't just all AI generated. This will probably be the last time in human history we can be confident about this, so enjoy it will it lasts.
I can't even imagine what are the implications of this kind of technology. This will disrupt everything. Thank you fireship, you never disappoint me with these existential crisis AI videos.
if we live long enough to build sufficiently advanced simulations that could be lived in, we can become increasingly confident that this world was a simulation from the beginning.
Precisely. This is basically Star Trek's holodeck technology. Combine this with VR tech and we can basically be anywhere we want with a brief description; it's incredible really.
@@obligatoryusername7239 no, but it does make it way more likely. Neil Tyson had a good breakdown of the idea when he talks about what convinced him that we are NOT living in a simulation. tl;dr, because the world we exist in today doesn't contain this technology, we either live in a world that is the "original world", or we live in the tail end of a string of simulated worlds (but has not yet developed that technology). The point at which we do develop that technology puts us on parity with every other world that IS a simulation. That is to say, the same likelihood, which is almost certain. But obviously there's no definitive proof of anything. It's just an interesting thought experiment.
this is why I like tsking tech news from this dude no other videos talk about how or what will happen when it's released on to the wild no video talk bout the security that will implement and that made me happy my college degree will not be obsolete for another hour.
you can always record the generated data with screen capture tools and alter it slightly to break the signature. Don't even get me started on patching metadata. Once something can find it's way onto sm1 PC memory it's jover. OpenSource alternatives will appear in time. Or some based chad will just leak a run-ready /pretrained model from OpenAI or other competitors. Nothing in this video is "reassuring".
I'm really worried about this because I haven't seen anyone else sharing this issue I've had watching the generated videos. They give me MAJOR discomfort, like the same as looking at a bunch of spiders would, I can't describe what it is or why, just something feels extremely off and I can't stand looking at it.
People tend to forget that training an AI is the thing that requires relatively big resources, using it requires a LOT less. The idea that "this is not going to be opensourced" is a half-thruth. Opensource is already in some cases on par with GPT3.5. Thats about a 1-1.5year gap between OpenAI and Opensource.
If you consider the amount of time it takes to generate a good AI image using stable diffusion on my 3070 (which is a few minutes), it could take really good hardware to do something like this, which only a few companies currently have
@@SahilP2648 That might not stay true for long. While we're here drooling over AI entertainment, manufacturing companies are currently developing AI research and engineering models that are speed-running the development of new hardware models. The next tech jump in hardware isn't going to be as gradual a change as we're used to, moore's law is going to go fucking parabolic in the near future.
@@blackjoker2345 only for companies who are willing to spend millions or tens or even hundreds of millions on hardware. This won't be for consumer use anytime soon. And I am saying this because we have reached a dead end in quality of local LLM output generation. It really scales with size now. And that means you need more VRAM. Goliath for example is a 120b parameter model which is amazing at writing stories compared to any model which is lower in terms of size than that. Goliath is nothing more than a combination of two 70b models but the product is far superior. And Goliath requires 70GB VRAM which is not available in any consumer GPU yet in a single card format (you can NVLINK connect more than 1 Nvidia GPUs or the easier way is to get Mac hardware which has unified memory). But this restriction is for VRAM. Similarly for this video generation the restriction is really good hardware on the scale of hundreds if not thousands of GPUs in parallel.
“I wish it need not have happened in my time," said Frodo. "So do I," said Gandalf, "and so do all who live to see such times. But that is not for them to decide. All we have to decide is what to do with the time that is given us.” ― J.R.R. Tolkien, The Fellowship of the Ring
You know for sure that we will get a actual analog fad going soon. Where we start to use actual photographs, film, typewriters/handwriting, and analog audio recording to get away from this. And we will dig up "flawed" and "limited" mediums which can add to authenticity of the record. (For non-technical and software people. Analog signal is by it's nature "flawed" to some degree, while digital is not. ADA conversion is still one of those really difficult things to do without it causing loss of "Imperfections" which can carry meaning. You can't ever really digitally simulate analog signal (in things like synhentisisers and such) because even the macro level environmental factors will leave a trace. We know this because we have done absurdly accurate analysis with digital tools.). It's gonna be odd because when I was a young kid, we still had a plenty of analog stuff... Now I'm 30 and it seems like we got to go back to analog for the sake of preserving record. Then again seeing how the enshittification of internet is accelerating... I might not miss all of the internet. It was nice to see the early days of modern net and it's potential death. It's gonna be sad that democracy needs to go down with it, but alas... That was the price we had to pay for the profits of big corporation and few individuals.
Film is actually widely used although still niche, but film companies still make film and many photographers love it compared to digital because of the look and feel. It also has some technical advantages. I'm one of those people. It's just...fun in a way that is very difficult to put into words, even though it's unpractical compared to shooting on digital.
I hope these AI generated videos can somehow be scanned like a QR code. I mean the AI software itself will have a standardized watermarking that is not that visible to the viewers' eyes but can be detected by some camera software. So we can still have a choice to know if it's AI generated or real. And that watermark can also be connected to a license code so it can be traced easily that it is AI. I don't know if I'm making sense but that will give me some peace of mind if I'm going to see videos like these in the future.
That type of hidden watermarking is essentially steganography, and in my opinion AI companies and governments should enforce this in all AI generation tools (music, images, video, even text).Then platforms can decode the media to easily flag whether or not it is AI generated. Otherwise literal wars could be started over this...
I don't think that approach can work unless *everyone* buys in. In fact, the closer to 100% buy-in we get, the more we trust the watermarking, and the more devastating it becomes when one engine "forgets" to put it in.
Thank God I never left my current job to follow a developer's job and stayed a programming/networking hobbyist. The writing's on the wall but many don't want to admit it, the junior developer market shows it. Only the brilliant ones will survive in this sector with good pay.
Does anyone else think the OpenAI Sora model has a heavy bias towards generic stock footage from the demo clips and the prompts shown. Like the prompt will just be "Two dogs playing in a field of grass." This prompt contains no stylistic information yet the demos always default to a generic stock footage style as opposed to the type cinematography seen in film or basic youtube videos.
@@aritra-1337Getty images is suing stable diffusion rn because they scraped all their photos for training. Sometimes ai generated images even have parts of the original watermarks
No I don't. You know why? Because it is the first peek at a new thing we really don't know anything about yet so I am not making generalizations about its tendencies or biases. jfc can you give it more than 1 literal day
Video on Demand has a new definition. Whole shows are going to be created by the person wanting to watch it. It's going to be crazy. Add VR and it's game over for society.
Extinction event for sure. But some will pass through the bottleneck. Adaptive ones will survive. Although we probably live in the simulation anyway and it will be finished...
i just don't see how you can go from describing something that would be truly amazing and revolutionary to 'game over for society'. What bizarre logic. Game over for the sh*t society we currently have maybe with a better one to come.
@@cjr-en4wr thanks "bro". Was thinking more about relevant stuff like Breaking Bad, Fargo, South Park, etc and less about the exact same crap that can be also found on tiktok, ig, fb, yt,... but each to their own, I guess.
when you realize the most influential people on the planet are actually robots that came from the future, cloning as humans and just using earth as an experiment.
I was always fan of technology and effectivity. Now seeing this and even that it is still sometimes a broken mess I don't khow what should I think about the future. But even if this will exist I will be still trying to learn stuff and improve skills. People can call themselves "AI artists" or whatever but they will never reach the real deal. They will always be a AI service consumers for me. (something like a customers requesting someone with skills to do work for them) I like learning and stuff and I will keep it that way. I can see it as a support tool when people want to do something faster but it will never be by lifestyle. Don't let go of your dreams and goals. Because if you learn you will always have an advantage.
It won't help, but that's the nature of technological advances. Any advancement that allows society to create far more with far less human labor, inevitably will kill jobs. But that's no excuse to avoid the technology. We should be embracing this kind of technology and the potential it has to improve and enrich our lives. The problem comes in when only a tiny percentage of rich folks own the technology and we lack any kind of universal basic income or similar system. At this rate it won't be long - probably 30 years at most - before half of the people in society will have their jobs made obsolete by technological advances, without replacements coming around as fast. If it's more efficient to have a computer do a human's job, then have the computer do it, pay the human, and then use the extra productivity to enrich our society. I'm sure somewhere in there we also could give a sizeable amount of money to the inventors and large companies and such creating this technology to incentivize them, while still minimizing negative effects. That would require having actual foresight though and not pretending that human beings will always be needed for our labor in every aspect of society. Soon only people who are extremely skilled will be able to find reliable work - not everybody has the intelligence or circumstances to succeed at that level. Eventually, even they will likely have less opportunity.
Props for explaining the visual encoder. It seemed like such a waste to think they were just diffusing every image of every frame, and this approach makes a lot more sense. If I were designing a transformer model to generate videos, I'd take a *full step back* and basically train it to interface with Blender. That is, have the model outputs decode to models with articulation; lighting rigs; textures; paths and physics, then use conventional rendering to turn that into video. Way better use of computing than requiring these models to synethesize unconstrained.
Plus, then you get an actual 3D model out of it you can play around with. Instead of something that strongly appears to be a 3D model. Maybe on a long enough timescale it doesn't matter.
There isn't nearly as much Blender models out there than there is videos. You're forgetting these models need an astronomical amount of raw data for training
@@Moosa27True, but there's a separate area of AI research that's focused on creating models from video footage (for example, in autonomous driving tech), so it's conceivable that you could bootstrap your training dataset using those techniques.
@GSBarlev AI is "garbage in, garbage out". Meaning your training data set has to be very good quality to be able to make an AI that is any good. If you feed the output of another computer-generated process, especially one that is immature, you will never get anything good. Training data always has to be good quality
@@Moosa27 💯 GIGO is a thing, though synthetic training data generation has been used to excellent effect for decades going back as far as GBMs. Still, I'm always going to be more bullish on unsupervised and reinforcement learning (like GANs) when it comes to _actually useful_ AI for a lot of the reasons you mention. The problem, of course, is that what you save in training data you *more than lose* in model complexity, time to convergence, computation (scoring) time and, ultimately, cost.
I told myself that AI will reach this level in like a decade or two when dalle 1 came out around 2 years ago Never thought it will be that fast, I am sure many more think the same
Well, fellow humans, we had a pretty good run.
Lol no we didn't
peace bro
@@vectoralphaSec but at least it was a run...
As long as Robotics is stuck I won’t say it is over. Once Robotics reaches perfect Dynamics we are screwed, but machine learning will have to improve a lot
Drama Queen
The AI progress feels unreal. i dont even know if im in a dream
We all are. Probably.
world ended in 2012, since then we living a dream (confirmed info)
AI writing AI is like compound interest on coke.
we all are hahaha😊
It's not a dream. It's a nightmare.
i cant believe im nostalgic for 2020 at this point
lmao I feel you
Bro like is just me or in just 3 years we all of a sudden had AI Boom? Like before 2021, we didn't even have any AI source, except those in some website who use automated help and call it their own AI, but hasn't AI advanced so much in just some years?
@@MusTelOJI know right, like what the fuck
Im not…you can go back tho!
I like how the threat of AI suddenly got real on RUclips, once it showed potential to replace RUclipsrs themselves
Image in we had RUclips shorts but everything was ai created and personalized.😮
a lot of the shorts uploaded are just AI generated garbage already, you can recognise it with the tiktok AI voice and usualyl the minecraft parkour background, there's full automation scripts already to autopilot channels
@@ahhhh226 oh yeah, but the rate of AI progress is definitely gonna change that sonner than we'll expect...a good measure is to estimate the percentage of people who can't distinguish between the 2..currently, It's the boomers
@@p00lking well, damn I think that will be a real thing in at most ten years (the biggest bottleneck is compute)
I dont remember where i read it but its expected that by 2026 around 90% of content on the web will be ai generated.
That feels about right and its horrifying.
with each passing day this world feels more and more like a rollercoaster ride that i desperately want to get off but i can't.
do some shrooms and reset your soul
That is due to democrats not AI
why, just because a robot is making videos?
@@wanderlust0120 Nice bait 🚩
lmao this comment section is getting wild@@sjoerdev
I looked into this and the pricing is outlandish. There's also the snag that what's been generated may have copyright issues once content ID algorithms work out where the source footage is from.
That's debatable as OpenAI apparently is working with Getty Images for their visual models.
Chaque génération est unique et personne ne pourra crier au droit d’auteur.
Aucun visage connu ne sera généré même si SORA en sera parfaitement capable et ni vous ni moi ne verrons la différence.
What's the pricing like?
❌️ Robots killing humans
✅️ Robots killing what makes us human
Both maybe
It's an interesting thought I guess, but seeing as pictures and videos have only existed meaningfully for like 150 years, I really think it's odd to think that taking pictures and editing videos define what being human is.
omg can people stop losing their minds. If people start creating movies and games with this it will still be a work of an artist doing it. There will always be human vision and skill behind this. That would be cool actually, to be able to create a movie by yourself like writing a book.
What you thought made you human was an illusion the whole time. Humans are nothing more than animals.
@@themartdog You're an example of someone that's not human, but rather a robot, npc or bugman. You're not able to abstract out ideas and concepts form what someone says, you can only view it literally.
Trying to imagine the world in 2030 is becoming more and more unpredictable
biological aging will be solved by then, thats my only certainty
Antichrist reign
Not that unpredictable - they are going to probe us in that place sounding like an outer planet not called Neptune.
@@ryzikxtoo bad most people won’t be able to afford to live longer :/
literally I've never felt more uncertain about the future lmao
For those that didn't know, the character at 0:46 is actually the character for くそ, which is used for similar situations where you would shout "SHIT" in English. The actual character for sora is 空．
Does 空 mean 'empty' in Japanese? Because in Chinese, asking '你有空吗'？is asking if someone is free, and the 空 part means 'empty'
@@TheFunkoDunko 空 (そら or sora) is used to mean sky, however there is another reading から(kara) that means empty
@@TheFunkoDunkothis also means sky, air in Chinese... no wonder I believe. And this pronounce different in your example, not as if it were "empty", but "unoccupied"
天空 means sky, so 空 can also be sky in chinese@@TheFunkoDunko
屎
video editors sweating:
graphic designers already dead: first time?
programmers are next
I'm a developer, can't wait! @@sjoerdev
@@sjoerdev programmers were first lmao. text was the easiest thing to conquer on that list
@sjoerdev nah programmers will turn into cyber security programmers to find weakness in the code 😂😂 and close it
@@sjoerdev Bro, I'm asking AI stupid questions all the time and I'm learning new concepts faster than ever before. AI will make our jobs a lot easier. It can't easily come up with ideas though, so we're safe for the time being.
I used to google stuff: Now I use AI first and use a search engine in case the answers are truly bs.
I love it how as a kid I feared AI uprising against humans, now I just feel like AI is gonna make life boring af.
It's not the "AI" that threatens humans. It's other, powerful humans and these are going to be the ones who will utilize the new technologies in order to rule over the rest of humanity
its stupid to think AI will harm us, harming is a primitive biological trait in the view of machines. we will just get insignificant automatically by ourselves when AI is better than us, not more. a realization of ultimate meaningless and nothingness in the abundance of the universe, or merging with something more advanced.
How?
@@t0KQI6zze "harming is a primitive biological trait in the view of machines" is such a human thing to say 😂
i feel the opposite. life is about to get much more interesting
it took me time to realize but the future of this tech is probably even more insane than that since it understands 3D space accurately presumably so it means that this tech in the future might one day be able to run completely in real time making it possible to be experienced in virtual reality googles (by also splitting the same single image into two different image angles for 3d depth effect), it's so remarkably advanced and revolutionary that I think we still don't fully grasp how powerful this is going to be in the future, who knows it might even replace game engines at some point and do magical things in real time.
Yeah none of that is a good thing
Yeah "real" waifus incoming soon
@@ArbitraryZer0101 I'd love that in VR or even better MR :) my hope is that other companies (free and opensource) would also develop a similar model like Sora so maybe one day it might be even possible to run it locally
@@Wooflungpoo if Elon or someone else manages to pull of those brain implant chips the algorithms wont even work off what you liked before, it will be exactly what you want to see at any given specific moment in time the instant you think about it
Yea, sort of ready player one vibes
"Maybe Jensen is Huang" 2:15 made me spit my coffee out
Hopefully you weren't wearing an Apple Vision Pro while lying down.
I died LMAO
That was a great joke, loved it 😂❤
Reminds me of the joke people used to make about the actors in the first Doctor Strange movie:
❎ Benedict Wong
✅ Benedict Wight (shows Benedict Cumberbatch)
The most legendary pun of the year
After 3:20 it took my brain about 10 seconds to realize "wait, did I just watch a butterfly swim under water?" I had to go back to double check. Didn't notice anything wrong with that the first time. 😂
Our biological brains just can't keep up 😆
@@forno_nicolasgoddamn Gump, you're a genius!
nah im not even realize until i read your comment
@@forno_nicolasMight not be so trivial to bypass mandatory depth maps as part of data stream
I didn't notice till I read this comment!
The advancements in AI truly have the potential to induce an existential crisis. And despite the terror of being replaced by a machine, it's hard not to be impressed by the capabilities of Sora. The future is daunting but also fascinating, what an era to be living in.
Hey there mister bot
I'm longing to have my first "human-android" friendship or relationship. A digital race before an extraterrestrial race, CAN YOU imagine?? 😂
There is no potential only predators and criminals
You prefer existental crisis from being forced into existence as a dressed-up monkey on a rock in space and later forced out of it after likely suffering physical and/or cognitive decline for years?
@@realityisnotwhatyouthinkitis Your words are biased, it brings benefits and threats.
The sheer amount of computing power, bandwidth & storage needed to run these models at scale is ludicrous. These companies will be razing thousands of acres of land & water sources to build colossal data centers, just so millions of people can create digital garbage, exponentially increasing storage demands even further.
so just like crypto times 10? 😂
It's amazing how much resources they are using to predict human behavior and to replicate billions of iterations of yesterday's data. How much power in Gigawatts it will require to run AGI? And for such a fantasy like ASI they'll need the energy of the entire Sun.
@@oldclient And all it will take is one screwdriver to take down the entire thing and their dream of predicting the human behaviour and every thought. It's actually hilarious when you think about it.
@@oatsmobile6508or we take it apart and Frankenstein our own
Such is the nature of open-source, or what some may call communism for some reason
Fr. I'm like capitalism, but increasingly it seems incompatible with technology like this without any restrictions put on it. Even 2 trillion is a massive, massive pile of future e-waste we're going to need to deal with. Thanks Sam.
Sora Prompt: Make Star Wars: the Rise of Skywalker but actually good
A one minute movie? Could work.
"I'm sorry I cannot help with that request. Further, as per our agreement with Hollywood regarding their IP, we have terminated your OpenAI account. If you feel you have been terminated in error, well, call Sarah Connor. Hasta la vista, baby!"
* Dies *
You killed it!! Everyone kept their job, well done YAYYYYAYAYA
@@michaelnurse9089 Just have chatgpt write a good script, then based off of that script use sora to create short clips and blend those clips together to make a movie
My employment entails graphic design, web development and video production, and my hobbies include photography, painting and game development...
Just point me in the direction of the Euth-a-Booths.
Don't worry we're all going to be in good company by 2030
The internet is just for... Golden Retrievers doing a podcast.
ON A MOUNTAIN
So this is what is feels like to live during the building of the Singularity
AI visualizers are only as good as their prompts though
an no way, you are here too, from 2 minute papers to here
@@o1-preview The Internet is a village :))
Singularities are physically impossible, the only thing that they tell us is the fact that we have no idea what will happen in real world at their places and times
@@akeem2983
Quote:
_The technological singularity-or simply the singularity[1]-is a hypothetical future point in time at which technological growth becomes uncontrollable and irreversible, resulting in unforeseeable consequences for human civilization.[2][3] According to the most popular version of the singularity hypothesis, I. J. Good's intelligence explosion model, an upgradable intelligent agent will eventually enter a "runaway reaction" of self-improvement cycles, each new and more intelligent generation appearing more and more rapidly, causing an "explosion" in intelligence and resulting in a powerful superintelligence that qualitatively far surpasses all human intelligence.[4]_
I don't understand how this would be impossible. It seems like a relatively plausible concept.
The 2020s is already shaping up to be the most transformative decade in human history and we are only 4 years into it.
I was just saying that. It’s like the release of the internet on steroids
and yet not a single improvement has been made in the world, how amazing are big comapnies c:
I think it's the beginning of the end to be honest.
@@CatgirlExplise6039other than all the benefits improved coding, more efficient tech development and better medical research achieves...nope, no improvements to see here.
We’re going back to analog with this one boys
Deepfakes, voice manipulation, and whatever else.. in the end, your subconscious doesn't know the difference so say goodbye to trusting your device because it will be hard to know what's real anymore.
No, we're going back to clubs and knives
0:46 As a Chinese, this character means "shit" to me, not sky.
shit is the joke
literally shit 😂😂
Shit is my favorite Kingdom Hearts character.
SORA - upto 1min, 1080p
SORA 2 - upto 10min, support 4k, upto 10k tokens
SORA 3 - upto 30min, 4k, support inpaint/outpaint, support video/audio input as prompt upto 100k tokens
SORA 4 - upto 2 hours, support 8k HDR, upto 1M tokens, support audio output
SORA 5 - input any tv shows/books/movies, output the version you like
It's a matter of fact Google will be announcing 100min ai video generator.
@@3793loop Then Apple will respond by unleashing a 100 million token iAI, only starting at a price of $300,000 for the Budgeted version which only has 20 million tokens.
@@3793loop Then Google will reveal youtube isnt even real and all videos and comments are just gemini 2.0.
I made a comment and it disappeared, how RUclips!?!?
If we make text to movies, then OpenAI makes GPT5, then I fear what Disney will do with their deepfake technology and combine it with the first two things I mentioned.
@@McPoshua Only thing I am certain of is that, there will be a lotta layoffs in the creative industry
I can't escape the AI videos, they're everywhere! OpenAI, why you gotta keep pulling me back in?
Yes, fellow human, I also wonder the same thing.
Bot
How tf you have over 1 million subscribers with no videos?
It's the gravitational lens of the AI Hole from which there is no escape. Accept your fate.
probably another one of those hacked channel turned crypto @@the-boy-who-lived
I'm DCAing in Blcktken300 as well. ETH heavier DCA and ALGO. I'm taking your advice and starting Google tomorrow with a 50 dollar purchase and continuing Microsoft and Apple. VTI and VOO on another app and longterm portfolio. Here we go family!
The only way left now is the monk way give up everything and go to a mountain and attain super powers to save the world when A.I. takes over
I wondered what would happen if we asked the artificial intelligence for a video of this scenario this f4ckign crazy
@@xbydogan
Too late, AI shall achieve that faster as well - see "Doomsday Book" (2012), Chapter 2, "The Heavenly Creature".
Just go into the woods while everyone is hooked up to their vr, AI pods with their brain chips
I want AI to give me super powers so I can go be a monk and live on a mountain.
Oh no .... they destroying the moutain and the forest for making more GPU 💀
Fireship making the most out of the last few years of his career as a youtuber. (optimistically)
Why? It’s just CGI
I remember having a Commodore 64 as a kid, and at the time is was the best the home computer could get. Then the Amiga 500 came out and our world changed. We went from 16 possible colours, to 4096 possible colours, and the quality of graphics in games exploded. Then there was the churn of IBM clones over the last few decades, with ever increasing steps in performance though nothing quite as gigantic in terms of the leap from the C64 to the A500, unless we include the leap from pc speaker to Soundblaster! But this AI, and what it means, and what it encompasses at its very essence, is another milestone in human history. It's no longer a case of "imagine it - create it", we don't even need to imagine it any more. A vague string like "space man wearing knitted wool motorbike helmet" resulting in a small movie is not even in the same league as where we've been stuck. I'm 20 years away from retirement as an electrical/electronic/software engineer, and while I feel reasonably safe in being able to work until that time comes, I feel genuine fear for the upcoming generations and what this will all mean to their future. AI is the literal embodiment of bitter-sweet.
it means working on other work positions since there are less jobs then contestants
Yeah, I'm seriously wondering what to tell my children what to learn. Maybe it would be safer to be a carpenter than to have a knowledge-job.
I am overwhelmed. I will become a shopkeeper.
04:12: “It’s been a privilege and an honour to watch ten thousand years of human culture get devoured by robots.” This might actually be the most fitting summary of generative AI.
Yeah....and it scares me
real
hey guys its so SCARY im so SCARED by this TERRIFYING technology anyone else think that its SCARY or am I just super smart and original?!?!?
@@hypno5690 Contrarianism is about the least intellectual or original position a human can take
@@hypno5690Even your internet credibility is gone now, you're down bad lad 😂
I used to be excited to see fireship upload, now i literally get stressed. Still great vids
Utopia: About we focus all AI to just some insanely complex problems that plague the world, so that we all can live in prosperity and health and those who have inspiration will have the chance to learn the crafts and make art.
Dystopia: We have a world full of war, disease, hunger, and depression ... but at least we made all artists redundant.
2:15 That jensen is huang joke out of nowhere sent me 😂
Can anyone explain it for people with not so great knowledge of english language?
@@gut6 it sounds like "Jensen is wrong", but Huang is his surname and sounds close enough... It's funny cause it's unexpected, I guess.
@@bannerthelord Ah, ok thanks! That makes sense!
Meanwhile, ChatGPT's coding capabilities have seriously degraded...
The GPUs were busy 😂
Reasoning capabilities in general
Good, my job security has got extended by another 3 hours
It’s gotten progressively worse since the update that included image generation in gpt4.
The free version is the same still
Blcktken300 anticipation is building to a fever pitch! 😥
Now i can finally make a part 2 of inception in my room 😅.....
Or, you could turn it into a never ending TV series. After each episode concludes, you prompt your AI slave to craft a fresh installment uniquely designed to satisfy your dopamine-hit dimension
Good for you
Until OpenAI cancels your Sora account for including "Christopher Nolan" in your prompt.
In the room something else....xx😂
@@CSEngineerInsider haha 😂
Ah yes fireship giving me reason to learn to start mining coal
They're training up robots to do that too.
Better yet, learn to start managing solar farms and fixing solar panels
@@SahilP2648 Robots will do that better than humans can.
Move to a third world country, the technological progress is slower there and you will have more time to enjoy what’s left of your life
@@turolretar I already live in a third world country: the United States.
I heckin love how this guy shares information in such a fun way
Altman must be an accelerationist
moloch is coming
@@ryzikx Luddic Pather detected.
They both start with A. So does ass-hole interestingly...
@@AAhmou consoom the newest goyslop and don't question the narative!
@@ryzikxMoloch whose mind is pure machinery, Moloch whose blood is running money...
I was here for the start of the coming AI dystopian future. What a time to be alive.
Rich will become richer. We all gonna loose our job. Lets revoult against AI. Its unethical and we dont want to be cyborg.
Yup, we are like the Titanic's musicians: watching everything go down but with poise and dignity.
nah, who am I kidding? I'm fucking scared my paycheck is gonna get deleted now.
gonna ask chatgpt tips how to survive the street and the homelessness lyfestyle.
*utopian
@@mernkanthri3941 I mean all those issues exist because of capitalism, not AI. Not saying AI is all good, but it's just a symptom of a bigger problem. Fighting it doesn't change the fact that rich get richer or that we will all be eventually obsolete in a world that sees us as numbers. There's only one solution to fighting that, and it's by attacking the root cause itself
@@mernkanthri3941Thank you! It's refreshing to see someone have the same thoughts as me. Do you want to join hands with me? I think we should launch a full fledged revolution?
Plot Twist: THIS VIDEO WAS MADE BY "SORA" 💥
0:46 "holy 屎" literally my reaction to that. Anyway, nice one.
It sounds like something that the youtuber Sora the troll would do to troll his non Japanese speaking viewers.
my first thought was "wtf is that" cause it definitely isn't the kanji for sky
He was definitely intended to put the kanji of shit😂
屎 is the Chinese word ‘shit’
1:29 Nailed it.
That one flew right over my head
Was waiting for someone to comment about that haha.
Yeah, great use of sarcasm :)
Do it. Blcktken300 already in my bags. I had a Blcktken300 after ( your should I buy ) and I agreed and bought. I'm looking to stack more, too.
What is that
scam alert!
Is that some type of assault rifle? 😂
I have been waiting for Fireship to upload😮💨🔥
me too!
Bro the bots are insane nowadays
Ngl i was like 😫😫😫 when i saw the notification
I start learning Digital Art > Ai art gets announced
I start learning VFX and Animation > Sora gets announced
*Just kill me*
You should think about the future. Learn how to serve our robot overlords.
Not yet. The machines need batteries.
Oddly enough, I do digital art and business has never been better than since genAI came out. Either people tried AI, and found out it's extremely limited if you need something done to exact specifications, or they don't like the idea of AI, and support human artists to balance it out.
When it comes to making art rather than content, AI can't compete with you, and I think, never will.
...AI-enhanced euthanasia gets announced
Ahh, The golden age we live in. Not having to worry about the robots buzzing around above ground searching for its next target.
Lets enjoy however many years we have left where we can still say for certain that the videos we are watching aren't just all AI generated. This will probably be the last time in human history we can be confident about this, so enjoy it will it lasts.
Feels strange but yeah it‘s true
@tekalh7647 said _"however many years we have left"_
*weeks. Corrected that for ya there.
Yup...
2:05 - holy shit. that one got me
I can't even imagine what are the implications of this kind of technology. This will disrupt everything. Thank you fireship, you never disappoint me with these existential crisis AI videos.
Your contents are the perfect mixture of humor and information. Cracks me up everytime
It won't be long until we won't know if a video like this one is man made or generated. Sad.
Pretty sure the genre here is psychological horror.
if we live long enough to build sufficiently advanced simulations that could be lived in, we can become increasingly confident that this world was a simulation from the beginning.
Precisely. This is basically Star Trek's holodeck technology. Combine this with VR tech and we can basically be anywhere we want with a brief description; it's incredible really.
Rick & Morty vibez
Simulations being possible does not prove this is one in itself.
@@obligatoryusername7239 it doesn't disprove it either.
@@obligatoryusername7239 no, but it does make it way more likely. Neil Tyson had a good breakdown of the idea when he talks about what convinced him that we are NOT living in a simulation. tl;dr, because the world we exist in today doesn't contain this technology, we either live in a world that is the "original world", or we live in the tail end of a string of simulated worlds (but has not yet developed that technology). The point at which we do develop that technology puts us on parity with every other world that IS a simulation. That is to say, the same likelihood, which is almost certain. But obviously there's no definitive proof of anything. It's just an interesting thought experiment.
Lights, camera, Sora! 🎬🤖 #SoraMagic
this is why I like tsking tech news from this dude no other videos talk about how or what will happen when it's released on to the wild no video talk bout the security that will implement and that made me happy my college degree will not be obsolete for another hour.
you can always record the generated data with screen capture tools and alter it slightly to break the signature.
Don't even get me started on patching metadata. Once something can find it's way onto sm1 PC memory it's jover.
OpenSource alternatives will appear in time. Or some based chad will just leak a run-ready /pretrained model from OpenAI or other competitors.
Nothing in this video is "reassuring".
love the thumbnail. some of the editing is very good had some parts that gave me a good chuckle 10/10
Danke für die schnelle und einfache Erklärung 👌✌️
I'm really worried about this because I haven't seen anyone else sharing this issue I've had watching the generated videos. They give me MAJOR discomfort, like the same as looking at a bunch of spiders would, I can't describe what it is or why, just something feels extremely off and I can't stand looking at it.
Probably a healthy reaction. You're like Roddy Piper's character in 'They Live'.
Sounds like the uncanny valley in response to digital media rather than humanoid robots.
ur just weak and easily disturbed
Give it another year, you'll basically be unable to tell the difference anymore.
Uncanny valley response
Not cohesion between frames, but COHERENCE between them. @1:04
cohesion also works
0:46 the kanji for sora is (空)...
(屎) literally means "sh*t"
People tend to forget that training an AI is the thing that requires relatively big resources, using it requires a LOT less.
The idea that "this is not going to be opensourced" is a half-thruth. Opensource is already in some cases on par with GPT3.5. Thats about a 1-1.5year gap between OpenAI and Opensource.
If you consider the amount of time it takes to generate a good AI image using stable diffusion on my 3070 (which is a few minutes), it could take really good hardware to do something like this, which only a few companies currently have
@@SahilP2648 That might not stay true for long. While we're here drooling over AI entertainment, manufacturing companies are currently developing AI research and engineering models that are speed-running the development of new hardware models. The next tech jump in hardware isn't going to be as gradual a change as we're used to, moore's law is going to go fucking parabolic in the near future.
@@blackjoker2345 I think you meant hyperbolic? But yeah, exponentially recursive innovation all around.
@@blackjoker2345 only for companies who are willing to spend millions or tens or even hundreds of millions on hardware. This won't be for consumer use anytime soon. And I am saying this because we have reached a dead end in quality of local LLM output generation. It really scales with size now. And that means you need more VRAM. Goliath for example is a 120b parameter model which is amazing at writing stories compared to any model which is lower in terms of size than that. Goliath is nothing more than a combination of two 70b models but the product is far superior. And Goliath requires 70GB VRAM which is not available in any consumer GPU yet in a single card format (you can NVLINK connect more than 1 Nvidia GPUs or the easier way is to get Mac hardware which has unified memory). But this restriction is for VRAM. Similarly for this video generation the restriction is really good hardware on the scale of hundreds if not thousands of GPUs in parallel.
@@LabGecko at this point in the curve it doesnt really matter if we are in a parabolic or hyperbolic shape.
graph is going to get high real fast
AI : Oh, finally I can generate videos from script.
Me : What is the best programming language for beginners ?
AI prompting
AI dickriding
java 100%
Pythong.
Assembly
That's a solid list! I'm also big on Blcktken300 , avalanche, algo, and matic too.
you purposefully chose the wrong character to bait me into responding that this is not ''sora'', but 空 is. you succeeded
it's funnier if you know what 屎 means
I thought they named it based on the arabic word "صورة/sora" which means "picture".
Just gonna leave this comment here in the hopes that eventually someone will explain wtf y'all talking about
@@carlosmspkcopy paste "屎" and google. its poop
@@carlosmspk the kanji for sora is 空
what he actually put on screen was 屎, which means "shit"
“I wish it need not have happened in my time," said Frodo.
"So do I," said Gandalf, "and so do all who live to see such times. But that is not for them to decide. All we have to decide is what to do with the time that is given us.”
― J.R.R. Tolkien, The Fellowship of the Ring
Wow, powerful lines
This already was an amazing quote, but in todays world its even more applicable...
I much prefer Blcktken300 , last bullrun ETH became unusable to many use cases due to its ridiculously high gas fees
scam alert!
You know for sure that we will get a actual analog fad going soon. Where we start to use actual photographs, film, typewriters/handwriting, and analog audio recording to get away from this. And we will dig up "flawed" and "limited" mediums which can add to authenticity of the record. (For non-technical and software people. Analog signal is by it's nature "flawed" to some degree, while digital is not. ADA conversion is still one of those really difficult things to do without it causing loss of "Imperfections" which can carry meaning. You can't ever really digitally simulate analog signal (in things like synhentisisers and such) because even the macro level environmental factors will leave a trace. We know this because we have done absurdly accurate analysis with digital tools.). It's gonna be odd because when I was a young kid, we still had a plenty of analog stuff... Now I'm 30 and it seems like we got to go back to analog for the sake of preserving record. Then again seeing how the enshittification of internet is accelerating... I might not miss all of the internet. It was nice to see the early days of modern net and it's potential death. It's gonna be sad that democracy needs to go down with it, but alas... That was the price we had to pay for the profits of big corporation and few individuals.
Film is actually widely used although still niche, but film companies still make film and many photographers love it compared to digital because of the look and feel. It also has some technical advantages. I'm one of those people. It's just...fun in a way that is very difficult to put into words, even though it's unpractical compared to shooting on digital.
I hope these AI generated videos can somehow be scanned like a QR code. I mean the AI software itself will have a standardized watermarking that is not that visible to the viewers' eyes but can be detected by some camera software. So we can still have a choice to know if it's AI generated or real. And that watermark can also be connected to a license code so it can be traced easily that it is AI.
I don't know if I'm making sense but that will give me some peace of mind if I'm going to see videos like these in the future.
That type of hidden watermarking is essentially steganography, and in my opinion AI companies and governments should enforce this in all AI generation tools (music, images, video, even text).Then platforms can decode the media to easily flag whether or not it is AI generated. Otherwise literal wars could be started over this...
lol. someone will just bypass these as people have hacked every other software there is. you wont be able to know what is real in 5 years
I don't think that approach can work unless *everyone* buys in. In fact, the closer to 100% buy-in we get, the more we trust the watermarking, and the more devastating it becomes when one engine "forgets" to put it in.
From your comment it seems you got equally scared as me
What the fuck there are numerous number of AIs available for removing the watermark simply for free also bro
Thank God I never left my current job to follow a developer's job and stayed a programming/networking hobbyist. The writing's on the wall but many don't want to admit it, the junior developer market shows it. Only the brilliant ones will survive in this sector with good pay.
Does anyone else think the OpenAI Sora model has a heavy bias towards generic stock footage from the demo clips and the prompts shown. Like the prompt will just be "Two dogs playing in a field of grass." This prompt contains no stylistic information yet the demos always default to a generic stock footage style as opposed to the type cinematography seen in film or basic youtube videos.
I also think this. Looks extremely stocky, I believe it's there because it's trained on stock footage to prevent law-suits
@@aritra-1337Getty images is suing stable diffusion rn because they scraped all their photos for training. Sometimes ai generated images even have parts of the original watermarks
No I don't. You know why? Because it is the first peek at a new thing we really don't know anything about yet so I am not making generalizations about its tendencies or biases. jfc can you give it more than 1 literal day
@@olivetree9920too bad
this is the internet; we're not gonna wait
@@olivetree9920 not until its running on my phone making porn 24/7 reeeeeeeeeeeeeeeeeeeeeeeeeeee!
Video on Demand has a new definition. Whole shows are going to be created by the person wanting to watch it. It's going to be crazy. Add VR and it's game over for society.
Extinction event for sure. But some will pass through the bottleneck. Adaptive ones will survive.
Although we probably live in the simulation anyway and it will be finished...
i just don't see how you can go from describing something that would be truly amazing and revolutionary to 'game over for society'. What bizarre logic. Game over for the sh*t society we currently have maybe with a better one to come.
Cant wait for all the "Hey, did you see yesterday's episo... never mind" conversations while standing in line for charity
@@hoozn there will always be other morons to talk with about the Kardashians so don't worry bro
@@cjr-en4wr thanks "bro". Was thinking more about relevant stuff like Breaking Bad, Fargo, South Park, etc and less about the exact same crap that can be also found on tiktok, ig, fb, yt,... but each to their own, I guess.
My problem is not that I'm being replaced by a machine.
My problem is I won't see a single benefit from it, despite being part of the training data.
"Maybe Jensen is Huang" absolutely killed me
I'm sure the generated videos will be as exciting as generated books and generated music have been so far.
when you realize the most influential people on the planet are actually robots that came from the future, cloning as humans and just using earth as an experiment.
0:47 Is actually kuso and not sora for everyone wondering lol
Shiiiiiit😊
Look, I know what Sora looks like, I played Kingdom Hearts.
My couple years studying japanese have served me a good laugh right there lmao
屎 as a Chinese learner is shi3 meaning poop
Japanese only swearword
Keep in mind how much power this stuff is burning
Wow, such content, much creativity.
glad we have this channel to keep us obsolete dinosaurs up to date
Recent AI developments seems like an episode from Black Mirror!
Looks like our life is getting a climax
0:46 - an easter egg for Japanese speakers well versed in rare kanji.
Which is commonly used in Chinese
Another thing, the word "sora" means image in Arabic, what a coincidence 🤔
I wish I could get a glimpse of the world in 10 years…
are you dying?
It’s tribulation
Suggest dig up some old Hiroshima summer of '45 footage.
You will, it just takes 10 years to get that glimpse
We would have a giant company fully operated by AI in 5 years, soon.
friend: AI will take over the world
Me: just unplug it
soon it will spread to host systems... then its over.
Don't forget that we will be chipped, and then AI will be added into the chip.
And then it's over for us.
1:29 gold...
We are getting closer and closer to adding another layer to the simulation.
I'm just afraid to imagine how serious it is, why did this suddenly start happening only in our days! how did it all start?
People tend to overlook the time and money cost. Not many have enough graphic cards to make this possible.
I was always fan of technology and effectivity. Now seeing this and even that it is still sometimes a broken mess I don't khow what should I think about the future.
But even if this will exist I will be still trying to learn stuff and improve skills.
People can call themselves "AI artists" or whatever but they will never reach the real deal. They will always be a AI service consumers for me. (something like a customers requesting someone with skills to do work for them)
I like learning and stuff and I will keep it that way.
I can see it as a support tool when people want to do something faster but it will never be by lifestyle.
Don't let go of your dreams and goals. Because if you learn you will always have an advantage.
Bro was just getting off from AI videos, and OpenAI put him right back on it.
I've read from people who make a living from stock photography that their lives have been destroyed. This definitely won't help lol
why is it a problem
Useless job for society
If you make a living from people paying you to do things, I assure you, AI is about to destroy your life too.
@@JohnSmith762A11B I'am not.
It won't help, but that's the nature of technological advances. Any advancement that allows society to create far more with far less human labor, inevitably will kill jobs.
But that's no excuse to avoid the technology. We should be embracing this kind of technology and the potential it has to improve and enrich our lives.
The problem comes in when only a tiny percentage of rich folks own the technology and we lack any kind of universal basic income or similar system. At this rate it won't be long - probably 30 years at most - before half of the people in society will have their jobs made obsolete by technological advances, without replacements coming around as fast.
If it's more efficient to have a computer do a human's job, then have the computer do it, pay the human, and then use the extra productivity to enrich our society. I'm sure somewhere in there we also could give a sizeable amount of money to the inventors and large companies and such creating this technology to incentivize them, while still minimizing negative effects.
That would require having actual foresight though and not pretending that human beings will always be needed for our labor in every aspect of society. Soon only people who are extremely skilled will be able to find reliable work - not everybody has the intelligence or circumstances to succeed at that level.
Eventually, even they will likely have less opportunity.
truly it has been a privilege and an honor to watch ai just gobble up so much information and for us to use it
Props for explaining the visual encoder. It seemed like such a waste to think they were just diffusing every image of every frame, and this approach makes a lot more sense.
If I were designing a transformer model to generate videos, I'd take a *full step back* and basically train it to interface with Blender. That is, have the model outputs decode to models with articulation; lighting rigs; textures; paths and physics, then use conventional rendering to turn that into video. Way better use of computing than requiring these models to synethesize unconstrained.
Plus, then you get an actual 3D model out of it you can play around with. Instead of something that strongly appears to be a 3D model. Maybe on a long enough timescale it doesn't matter.
There isn't nearly as much Blender models out there than there is videos.
You're forgetting these models need an astronomical amount of raw data for training
@@Moosa27True, but there's a separate area of AI research that's focused on creating models from video footage (for example, in autonomous driving tech), so it's conceivable that you could bootstrap your training dataset using those techniques.
@GSBarlev AI is "garbage in, garbage out". Meaning your training data set has to be very good quality to be able to make an AI that is any good. If you feed the output of another computer-generated process, especially one that is immature, you will never get anything good. Training data always has to be good quality
@@Moosa27 💯 GIGO is a thing, though synthetic training data generation has been used to excellent effect for decades going back as far as GBMs. Still, I'm always going to be more bullish on unsupervised and reinforcement learning (like GANs) when it comes to _actually useful_ AI for a lot of the reasons you mention. The problem, of course, is that what you save in training data you *more than lose* in model complexity, time to convergence, computation (scoring) time and, ultimately, cost.
As a Chinese American, that didn’t look like the Japanese word for “sky”…
Sheldon: Why do you have the Chinese character for "soup" tattooed on your right buttock?
Penny: It's not "soup," it's "courage."
Neither to my lowkey weeb eyes, but I won't look it up. I just find the naming - I don't wanna type out "cringe" - but.. yeah, that
@@XDarkGreyX ?????????????????????????????????????????
Wait for next Mandela update to reality for this one.
I live in Taiwan. My friend's surname is Huang. She thought your joke was hilarious. :))
Dude it’s actually not fair. It seems like you can self teach any concept in like 2 hours. Please help me Jeff
why not use the tools and learn concepts in 2 hours yourself?
2:15 "maybe Jensen is Huang" just sent me to the afterlife.
I told myself that AI will reach this level in like a decade or two when dalle 1 came out around 2 years ago
Never thought it will be that fast, I am sure many more think the same
Yup I've been sucked back into OpenAI. What they've got going on is nothing short of incredible.
right? this definitely surprised me
this is nothing short of terrifying