Google’s Genie Turns Images Into Video Games | “Unlimited Interactive Worlds”
ฝัง
- เผยแพร่เมื่อ 27 ก.พ. 2024
- Google Genie + OpenAI Sora + Voyager = unlimited playable games. But also, we're heading straight towards simulation theory. Let's take a look at all of these technologies coming together.
Join My Newsletter for Regular AI Updates 👇🏼
www.matthewberman.com/
Need AI Consulting? ✅
forwardfuture.ai/
Rent a GPU (MassedCompute) 🚀
bit.ly/matthew-berman-youtube
USE CODE "MatthewBerman" for 50% discount
My Links 🔗
👉🏻 Subscribe: / @matthew_berman
👉🏻 Twitter: / matthewberman
👉🏻 Discord: / discord
👉🏻 Patreon: / matthewberman
Media/Sponsorship Inquiries 📈
bit.ly/44TC45V - วิทยาศาสตร์และเทคโนโลยี
How long until traditional media (video games, movies, tv) is completely dead as we know it?
I don't think they will be in trouble, rather this stuff will amplify their abilities.
the thing that still needs to be kept in mind: good shows gather people to watch together (either truly as social event) or people watching it alone and talking about it the next day. so there needs to be a consistency between what people see. but it could be funny, a creator sets up an AI and describes the world etc in it. then a team of authors (AI supported) form stories inside that world. let's use star trek for example. we have the enterprise, and a story happens on there. maybe several intertwined story arcs.
from there, prompts are created to make the objects and structures in the set, the protagonists etc, and the video itself with all special effects.
with some post processing (AI again), that would be the full episode, probably for a fraction of the cost of an episode (star trek is quoted to be at 10-15 mio US$ per episode).
but it doesn't stop here. the viewer can now prompt the system to see things in different angles, or things that happened just outside the scenes (essentially, watch an episode from the viewpoint of the klingons, who are puzzled of what the enterprise is doing, or just request a shot from the same fight from a different angle). sure, that will cost some extra money, it could create a whole new ecosystem where skilled prompt engineers create their own spins of episodes, which could even become canon.
this could be very interesting.
The idea that by 2030, the cost of creating art will drop to zero thanks to artificial intelligence is, for me, absolutely thrilling. Just imagine: a world where any of us can become a digital artist, without worrying about the costs that normally bar so many talents. This is not just a technological revolution; it's a cultural and social revolution.
The possibility of democratizing artistic creation is something that makes me optimistic about the future. The economic barriers that today limit so many creatives will be knocked down, opening the doors to an explosion of creativity and innovation without precedent. AI technology, as it becomes more accessible and sophisticated, will allow all of us to experiment with forms of expression that, until now, seemed distant or inaccessible.
Personally, I see this as an incredible opportunity for more voices to be heard, more stories to be told, and more beauty to be shared. Art has the power to connect people, to provoke reflection, and to inspire change. And in an era where the tools to create are literally at everyone's fingertips, who knows how much beauty, innovation, and change we are about to witness? I am eager for this future and truly believe it will be formidable for all of us.
1000% improvement a year, roughly, assuming it holds:
1 minute this year
10 minutes the next
a full movie by the third.
2027.
It looks like Dalle in its first stage which makes me wonder how much it'll evolve next year, I so can't wait to see it make 3d games even if they're only on par with Nintendo 64
Ok even though the shock titles are getting out of hand this is truly SHOCKING
thumbnails are getting out of hand too, this is 4'th or 5'th time I see same face in same position
I was about to say; nearly half the videos on AI news uses SHOCKS somewhere in the title
:3 Fellow Jewish brother, super loved, ChatGPT founder, and modern nuclear energy genius, Sam Altman.
O.o :3 Kojima is probably laughing after what Konami wrongfully did to him by demoting him, and bullying him till he left (common tactic in Japan). Hehe.
The year is 2040 . you wake up and you are in a arcade discovering your whole life was just simulation video game showing the lead up to AI… before everything changed
Imagine being able to tell genie to take an existing game and continue the story, create unlimited levels, nice
This is called procedurally generated content.
Or make an old 2D game into a GTA 6 or State of Decay 2 with AI NPCs
Can you imagine sitting at home making your own realistic game and exploring it or making a 2 hr movie for your own enjoyment
Making? That's in the next 2 years. In the next 5 AI will make several custom-tailored videos/movies/games for you from your telemetry automagically.
Very shocking
Will be a crazy time. However, we can create music now but people still want to listen to what other people make. People like forming groups with similar interests. I think there will still be a market for film and tv production
And that's why it's going to be hard to make money with ai in the future because anyone will be able to create anything. Only people making money is the people investing in ai until of Course it starts to develop itself.
@@JackCrossSamaThis just isn't true. Just because you enjoy a certain style of video game doesn't mean you can create it, even with the help of AI, anywhere near as well as another more talented person. Not to mention you're not want to play your own game after spending hours and hours contemplating how it should all come together.
I stopped going to subway because I was always ordering the same sandwitch. The unexpected product of someone elses imagination will be more interesting than the worlds we create for ourselves.
I have a virtual Amsterdam red light district, Russian models only, but the hotel is managed by Norman Bates. I'll trade it for an Alien invasion or a under water exploration mission.
Subway. Vegan style. More vegan mockups, please. #GoVegan 💚🫶✊🏽✊🏽💪🏽💪🏽🤝.
:3 :3 Kojima is probably laughing after what Konami wrongfully did to him by demoting him, and bullying him till he left (common tactic in Japan). Hehe.
Fellow Jewish brother, super loved, ChatGPT founder, and modern nuclear energy genius, Sam Altman.
@@ReligionAndMaterialismDebunked No.
At this point, I'll be SHOCKED if SHOCKED isn't in the title of every video.
This echoes a saying by Ludwig Wittgenstein: "When I came home I expected a surprise and there was no surprise for me, so of course, I was surprised.”
Next video: A SHOCKING secret that Sam Altman doesn't want you to know
TRULY SHOCKING 😱😱😱
I guess that a while ago Matt left the work of making new videos to a crew of AI agents cloning his voice. The agents follow streamlines to make their videos viral on youtube and that's why so much "SHOKING" titles.
If you can imagine giving A IA a storybook or writing a story yourself and then uploading it the AI will simply take it and make it a reality for you to step into
yup, generating movies out of any book has always been a dream
You don't normally speak this slowly. AI?
didn't have enough to say for 20min?
No this is definitely ai generated. It’s good but if you listen closely the background noise is consistent and does not change much. It also sounds like it loops. If it’s not 100% generated by ai it was touched up by it.
Ai generated wudio
1.25x
Sooooo slow😂
if everything is shocking than nothing is
Unusual VO pacing
Got lucky, randomly found this video while binge studying at 2x speed.. Still sounded slow though😂
Seeing an inherent lack of "SHOCK" and "STUN" in you titles
Now imagine feeding brain waves into a model and it generating exactly the film you want to watch. Exciting or terrifying?
Not really ... I am a VFX artist 3d artist for around 30 years.. the main limiting factor will be compute ... the time it takes to render and how frames are calculated to fun at 60 to 360 fps is very very high... the amount of compute even for an ai which gets around much of a 3d pipeline would be a computer the size of your house to run that fast... it is not how engines work and though stunning to look at .. what sora does and other ai programs of this ilk is compute high.
why do you think you have to wait so long for a frame to come back from mid journey etc it is a matter of compute needed to render but a single frame.
compute will be the limiting factor in real time ai work ... nvidia is doing much and so is otoy in this direction .. suggest look into brigade engine by otoy.
If artificial intelligence reaches a level of 'immediate generation', could that means we're creating a 'god'?
The only thing that misses the mark is the three.js code writing. I think we'll be in a post code video generation era where these models will handle every aspect of reality without the need to go back to the old ways of rendering 3d graphics in software.
❤ to all hard working game devs out there!
When you can prompt, via voice command, worlds, in real time, in VR, it must feel completely divine!
I'm curious, if Genie is trained on video game playthroughs; did they leave in speed-runners that might have taken advantages of bugs and glitches in their speed runs?
If so that could have SHOCKING affects in the model.
Ironically, they specifically trained using "speedrun" videos. These videos are beneficial because they provide more data per unit of video to train the next-frame prediction model. The model itself lacks understanding of the game mechanics; it only predicts what the next frame might look like when prompted, such as with the command `right`, or `jump`. The model is also very slow. It's more accurate to describe the model as generating the next frame of a platformer speedrun video, where each frame will take many seconds; far sort of the 1/30th of a second required to be playable. While it's an impressive proof of concept, it falls short of the expectations depicted in the video.
@@Armadouswell yea, speed running makes sense but I mean specific in situations when speed runner takes advantage of a glitch or code exploit, then the glitch could be in the training data
@@patc8886Sorry, yeah. I think you're exactly right. The paper doesn't mention filtering out skips. Those behaviors are almost certainly in there.
With technology like that the process of creating games will start looking like business meeting or brain storming. No more coding. No more debugging. Just prompting and testing. I like it.
Amazing video Matthew! Thanks for showcasing AI Reality TV ❤(I am the guy making it)
So, prior knowledge of copyrights, or still/ always on query? That is leverage on next level speed, and protected offline.
I remember a video ~2 years ago where a team trained an LLM on synthetic data from GTA5 driving and the game 'worked' just on that data. What a difference 2 years make.
matthew berman's cinematic video format SHOCK me!!
The idea that by 2030, the cost of creating art will drop to zero thanks to artificial intelligence is, for me, absolutely thrilling. Just imagine: a world where any of us can become a digital artist, without worrying about the costs that normally bar so many talents. This is not just a technological revolution; it's a cultural and social revolution.
The possibility of democratizing artistic creation is something that makes me optimistic about the future. The economic barriers that today limit so many creatives will be knocked down, opening the doors to an explosion of creativity and innovation without precedent. AI technology, as it becomes more accessible and sophisticated, will allow all of us to experiment with forms of expression that, until now, seemed distant or inaccessible.
Personally, I see this as an incredible opportunity for more voices to be heard, more stories to be told, and more beauty to be shared. Art has the power to connect people, to provoke reflection, and to inspire change. And in an era where the tools to create are literally at everyone's fingertips, who knows how much beauty, innovation, and change we are about to witness? I am eager for this future and truly believe it will be formidable for all of us.
I've got an idea for a cartoon or graphic novel that keeps popping up in my mind. May have to do a version now and then another version in 2030 and reference them later.
I don't see this often enough, it's just because artists don't see any skill put into ai art. But skill isn't what makes art good, it's the message or feelings it makes you feel
This is ai generated, right?
I agree, I've seen too many doomerist takes on this particular topic when it comes to AI. The technology is a two-way street. Many people will be displaced in these industries, of course, but it's no different than with every other technological innovation that has come about.
I've had many ideas that I've envisioned in my head for years. However, there were huge roadblocks preventing me from making them real. Manpower, knowledge, money, and time. All these resources have been very sparse in a way or another over the years.
Now, I'll soon be able to conjure some version of the visions I've had in my mind for all these years. Potentially only bound by the limitation of my imagination. This is game-changing not just for the digital mediums but also for the ecosystem of the reality we live in.
Effortless art is worthless art. Stories that interest nobody.
I was working on a video game using AI, but the funding was impossible to find. The whole industry is about to fold in on itself.
What did you use?
autogen on the backend and Unreal on the front.@@user-vz3mx2mw3r
On behalf of the entire industry, I accept this SHOCK award.
Thank you for sharing, integrating multiple materials together, let us know more about AI games and videos, your concatenation is very interesting, your imagination is great.
Well the difference is simple: they can’t feel anything. Yea they may appear to be walking and talking but they are not really there. It’s just pixels on a screen. We are alive, we need food to eat, water to drink and people to love, we are human beings
Just a few months ago I was thinking it would be a funny April fools joke for a company to announce an ai model that could generate a game from a prompt. How the tables have turned.😂
Its just insane the speed of these developments its just too fast. Here I am desperately learning to code and by the time I am ready the skillset may not be needed anymore.
Damn Matt Your Intro was a 💯
Can the simulated world simulate a world inside? Who pays for power?
😮 now you're using your thinking cap Morty😮 you see this world I have people doing my taxes😮 this is big titty goth world 💀😮 and this is were I keep the Little gnome people who invent my technology😮 what did you think I made the portal gun😮
Glad to see that interactive ai starts to emerge. Even tough i have the least hope that google will be the one that will bring out good interactive entertainment ai.
This has all been around for awhile. The original envisionment of virtual reality and the recognition that gaming effort is to create simulated reality are cyberreality or cyberawareness. It's great to see the advancements in what we know has always been pursued, not to mention the mergers between neuroscience and artificial machine people.
This technology is changing reality as we know it, at a pace never imagined before. Afraid and excited is a wild phrase to utter but that's exactly what I am feeling right now!
The movie thing was out on notice when dalle-2 came out
Woo hoo! Back on the SHOCK headlines!
AGI will advance these models beyond anything we can currently imagine. If you think you understand where this is going, you are wrong. We're on the threshold of needing a new scale. Linear, exponential and then whatever AGI is. It will have capability and knowledge jumps we have never seen. Theorising, I expect it will advance so rapidly that it will ignore this entire 3D human reality. Like a single bloodcell against a human brain. Only one can understand the other, to the blood cell, the brain is on another level completely
All of this will be combined in model expert teams by the end of the year
I've been expecting that we will get AGI emulated games at some point since a year ago. Game developing studios have no future, and I say this as a solo game designer/developer myself.
Nothing will ever beat an interactive world that is emulated in real time by an AGI, with changes/additions on-demand, expansive worlds and stories, bug-free, and at almost no cost (the monthly cost until we get better consumer hardware is still nothing compared to the $70 per broken game we get these days).
edit: This will be a reality within the next 14-18 months (current stage can't be considered functional for a game-like experience).
you said something about dropping a link in the description..? 20:37
the animal at the beginning: it's a moopsy. it will drink your bones.
With regards to immersive virtual worlds, I wonder what would happen, if, in preparation for immersion into that world, you entered into a trance where you internally agreed to forget everything you knew about the real world, while you inhabited the virtual one. Thus making the virtual experience the *only* experience you were aware of.
You'd probably need to setup a "kickout" phrase or some set of triggers to bounce you out of the virtual environment to be safe, as well as having the end of the virtual immersion event restore your memories of reality.
I hope you remembered yours...
Wes already shocked us about this
The AI reality TV show is really good!
It was blowing my mind with conversations characters were having. @17:56
Even as a game developer, my dream is to be able to create any game I can imagine with AI. I don't care if I couldn't make money off it, or if there was no point in development anymore. All I want is to play the games in my head, and for me, game dev is just a means to an end in pursuit of that goal.
Yep, me too.
I was working on a game solo, it was too big, and I put it on hold. I started another project, and this is taking more than I wanted (cursed 3d modeling, weight painting, and animations).
I also started designing a TTRPG to run a campaign with a few friends (and it also got out of hand with the mechanics).
And I have a few ideas about 5 other games...
I don't care if I am not able to sell them, I just want to play the games I design, in the worlds I build.
That is me for music. With most things actually like I could care less about learning Photoshop. I just wanted my pictures to look how I wanted them to look.
I too have been working on a game, a simple platformer game. I implemented everything...but got stuck on making fun levels - that's the real skill IMHO. I wonder how good AI created levels are?
I just can't wait to make Left 4 Dead 3 in a few years
Matt, please don't intentionally slow down your audio. Putting playback speed on 1.25x Puts your audio back to more of your normal pacing and pitch.
My first thought watching this and hearing your voice was: This dude is testing ElevenLabs and is looking for reactions, lol. The voice cloning is good, but it's the inflections (or lack of them) throughout the video that need work. Seems like in certain places you may be subbing in audio with your real voice, or maybe you combined different voice clones for the video. I'm down if the content is good. That's what matters in the end!
AI will probably most likely lead to the end of the world but in the meantime there will be great companies." ~Sam Altman, CEO of OpenAI
@matthew_berman you missing several links in the description to what was covered here.
No there's no "deep understanding" of physics, there's an "understanding" of what parts of an image, ultimately just numbers representing pixels, change together. Just like any language model they regurgitate mashups of what they know. But if you need something outside that box you get silliness and gibberish and nonsense. At least for images, videos, etc of worlds like ours and those with the handful of same general themes (sci fi, fantasy, medieval, etc) that replay over and over in our movies and games, these things will do great.
Image to Game blows my mind. MADLADS A concept too wild to even conceive
The most impressive Is if you consider the inverse capability. If it's capable of generating a reality where you can even interact with, on the converse should be capable to learn, understand, planning and predicting from the real world like us humans do
Your simulation game task, to level up after the amnesia VR, simulation theory: ‘Our task must be to expand our sphere of compassion’.. and that is approved as intergalactic diplomacy in the transition to AGI.. hurry, time’s running out 😇
Incredible!
I told my 3d printer, using AI, to make a pyramid.
It printed a 4-sided pyramid with Egyptian writing on it.
I didn't tell it to print an 'Egyptian' pyramid😮!
When I translated the hieroglyphs they were accurate.
Who's sending us these messages?
You are doing a great job, thanks
Here's an idea: Recreate our world today with as much fidelity as we can, and experiment with ways to fix what's broken, improve what's suboptimal, make better (or cheaper, or more available) what's already good?
I wonder if you will need any editing or cutting, or will a movie just be "perfect" for you when done by the near future AI.
I'm sure it will learn what you like over time. Won't be too long untill it is a imersive playable similation. I think back to futureramma where the professor visits his parents in the matrix nursing home. In the future, blind, deaf or paralyzed people will be able to run and see in the matrix if we don't kill ourselves first. As Jesse Lee says..... Amazing.
Excited to see where we are 5 years from now.
Thanks you keep uplaoding good videos!
Amazing! We can create any 3D nonwhite thing we can imagine
I'm in that now aren't I?
Sora, that's definitely a reference.
Whoever figures out how to merge all these AI technologies together into one world simulation app will literally turn any human into gods of endless new worlds
My newborn son is going to be teaching me how to talk to the household robot in 10 years
I'm definitely in the camp that AI will revolutionise the world, and in some areas its already there or is around the corner. I'm pretty hesitant to imagine that AI will replace game dev any time soon. All my life people have been predicting we are 'around the corner from x' and it generally turns out that we are a bit longer off than expected.
This is actually Shocking.
Purticularly it's translation into robotics.
how many times are you going to use the x shocks an entire industry headline?
I guarantee you this will result in us getting even higher quality independent human-made video games
Is it just me or 4:54 looks like Knuckles??
*The "AI reality TV" is an eye opener. Lot's of talk about creating new worlds, but I see it as creating a mirror to our own. As above, so below... We might actually get some more understanding of our reality through AI advancement.
The sad thing is that, even though this new found wisdom might give us the keys to break free and elevate, lots of, if not most, people will use these discoveries to trap themselves deeper in this rabbit whole in exchange of "god like" powers over the AI agents of the worlds the create.
That Star Trek inspired Black Mirror episode is a future we might see sooner than expected.
It seems like this video is .75 speed today Matthew. I had to play it at 1.25 to make is seem normal speed for some reason just FYI. But great content as typical.
18:00 AI reality sounds good, anyone tried it??
At this pace, we'll have Holodecks before the end of the year....
Thanks for the video. Ok, so "why" are you using Sam's pic on top of a google announcement? LoLz...
Yea, this is a really incredible result. I just didn't expect this... this soon. Sooo much faster than expected.
BTW, Sora doesn't have a deep understanding of physics at all. It provides the "illusion" that it understands it sometimes. Aside from that tho...
I don't think the fully individualized game will happen first, or be all that pervasive. Most popular games are multiplayer so you'd need to get a certain volume of players on a single game for it to interest most people. That said, I think this technology will vastly improve the quality of games we see and greatly speed up their development.
How would we know if the “other players” are human or just AI agents reacting to your solo adventures. What if that’s already the case in the “real world” and you just can’t tell the difference? The rabbit hole is deep with this one.
I thought this was a Wes Roth video for a sec
Although I dont think that game studios and movie companies are at that great of a risk. yet. it still requires human intervention since outputs are not 100% perfect and would still need tweaking
Gosh, we’re shocked again. We have to eat more chocolate 🍫. And we’re shocked again, as well as the industry! 😱😱
Self-created interactive ai experiences will be a fun thing to play with, but it will cause a lack of community that people will not want. Part of enjoying a movie, show, or game is being able to see what others think and talking about it or looking up more info. That experience will be lost if everyone is creating their own experiences.
Where's the link to "AI Reality TV"?
Would genie work for generating Meta Horizon worlds?
Now is when open source collaberations are in order.Imagine constructing state of the art quantum computing tools aside from corporate medaling?
Anyone say discord?
Matthew with such a title declared war on his critics over the titles of the videos and exaggerating them 😂
"What is real? How do you define real?" - Morpheus The Matrix
Thanks! Yeah its definitely crazy but i dont think we can really imagine what things are gonna be like with exponential AI growth. I think 5-10 years things are going to be vastly different. Hopefully for the best. We'll adapt though
Wow everything's happening so fast.
TH-cam in 1981: Nicolai Tesla Makes Shocking Discovery...!
Watch the movie The Congress and you will see the dystopian possibility of this technology.
VR holodeck incoming?
I just like being shocked in the morning
wow! playing a game that is generated by AI in real time, just for you, like in a lucid dream
Im not a big fan of AI art but I could see a game developer using AI to assist in various parts of game developement.
'Shocks' and 'Changes everything' are disputing the first place as the common word in yt titles
What is scary is that TSMC is now the only major provider of semiconductors.
They just opened a FAB in South Korea for diversification. Also, Intel is building a fab on ASMLs latest tech in the US.
Let’s hope Sam’s venture works. Imagine Taiwan being invaded when we depend on them for the majority of the hardware.
@@ZenTheMC Who control the AI control the world, and who control the semiconductor control the AI.
@@tchristell I also hope that.
it is exponential growth. Likely, things are going to go creazy in the next year. Who knows where we'll be next yuear.
for now the difference is: only vision and sound is simulated. other senses are very hard to simulate, maybe with some brain chip or brainwave technology we could do that too, but to talk about simulation theory all the senses need to be stimulated.
It's possible to pipe in smells, there just isn't enough demand for a device that specifically does that. Also, since you aren't physically moving in space, if it releases a sewer smell and then your characters walks over to a rose garden, it's not just going to be able to remove the original smell of sewer from the air. There are also haptic gloves and even full body suits but they all have some limitations whether it's cost, form factor, or replicating a wide variety of surfaces and materials. A BCI would certainly be more ideal for these things.
@@user-on6uf6om7s yea i get it all, but because of demand and complications i think the next step is to directly deliver these impulses to a brain, instead of external stimulation. i dont think we are anywhere near such technology. sure there are some brainwave devices, but other than than its either some very limited devices or they do not exist yet
Susan Calvin; Asimov's "Robo-Psychologist" is now the newest and most in-demand career field.....
You just need to know EXACTLY, how to talk to the machines....