OpenAI's STUNNING OMNI MODEL | GPT-4o is being released into the wild...
HTML-код
- Опубликовано: 1 июн 2024
- Learn AI With Me:
www.skool.com/natural20/about
Join my community and classroom to learn AI and get ready for the new world.
#ai #openai #llm
TIMELINE:
00:00 Two GPTs Talking to Each Other
05:53 OpenAI Spring Update
06:27 New Demos
24:00 OpenAI's Spring Update Event
49:10 Ending Thoughts and Analysis
does this mean I can get my UBI check now?
Yes
Santa will bring it, be patient 🙂
Not yet but we are getting AI gfs soon
I want mines too!
Nope the money is going into making police robots.
I think its hillarious that the AIs don't care about what we typically care about and go off on a tangent talking about the lighting :P
it's basically an autistic person
That's being done strictly for user interface.
Babies are fascinated by these things e.g. colour etc
You’re running under an assumption that people don’t think lighting is interesting just because you don’t find it interesting
What did you find interesting about the room? Haha
Now all HRs are going to experience what a layoff feels like from the other side of the (desk) webcam
Why are there so many people that are still blind and refuse to acknowledge how good AI is becoming.
and how is that a bad thing? its evolving
I've been trying to show and tell people for the past 1 year and literally even people that understand technology still can't see how big this is.
@@faizywinkle42I didn't say it's a bad thing. But I think it's a bad thing how many people are clueless. I've been explaining to this one guy that an AI will be able to do his job and he's like "well AI cannot go back and forth through emails with the knowledge I have". I'm like bro AI can certainly do that. I think many people simply have no idea what AI is. Until they see robots in the street, they won't believe it.
It's denial at this point
People are deathly afraid of it! It was fine and cute when it was in Sci Fi movies but now that it is real some people's heads are exploding! Especially non-technical people, it is like evil magic to those people! To them this is almost as if Aliens landed on the White House lawn!
Literally samantha from HER
Definetely feels like an intentional nod😅
I would had say the same, but you were first here.
Now will that ends like Samantha's end of the story, i.e. leaving us, poor an so limited creatures to explore the universe in better company ...?
VERY much! Love it!🥰
Ew. Yuck. Give me the male version. ❤ 😍
SamA wrote a one word tweet earlier: “Her”
This is the first Wes Roth video for which I think "STUNNING" is actually an understatement
Complete opposite for me. "Stunning" is the world's biggest overstatement. Underwhelming is perhaps too generous of praise for this marketing hype. Sora was stunning. GPT4 was stunning. A desktop app isn't. A cheaper smaller model is nice but still isn't. And text-to-voice AIs have existed in the open source land for months.
@@thomassynths "Marketing hype"? Are you suggesting that the presentation is a distortion of the truth? Otherwise, I think you might have misunderstood the real nature of this breakthrough.
@@jason_v12345 I didn't say hoax. I said hype. Days of pre-advance hype... and we got this demo. Meanwhile Sora dropped out of nowhere and floored the world.
@@thomassynths same I thought, the whole thing is just not that impressive.
They only put existing things together and made it maybe a little faster, but with loss of quality coz the AI-voice sounds terrible. Like in the introduction video they only presenting very very very simple tasks, like solving 3x+1=4 ... so thats not enough to impress me.
Bringing all together is a step but not a big one.
@@thomassynths Well, maybe it's beacuse you didn't catched that it's not text-to-voice AI, but voice-to-voice one?
And people still believe their job is safe 😂😂😂
🥶
Do you think Congress will save them? Congress is a waste.
Maybe jobs are not what we think they are? Maybe it’s time we think beyond jobs and start looking at what it means to be human beyond the occupation! 🙏🏻
This is one of the most profound comments i have read in months about ai will change our lives and society @user-sl9mc5rw9g
@@user-sl9mc5rw9gThis!
google announced this. and failed. now closedai made it a reality
Business implications for OpenAI and Microsoft are massive. They made this model free to access without an account. That means that every person who interacts with it that doesn't specifically opt out will be providing multimodal training data for iterative improvement. And the ability to watch someone's screen and see what they do after the model tells them how to perform an operation or do some function means thousands of hours of training data on direct digital action over multistep workflows per day. And the same comes from their business partnerships (though not enterprise unless the company opts in). No wonder they aren't worried about not having access to any private training sets. They are never going to run out of training data again.
Data scarcity is a total myth anyway. There is data everywhere, and it is being created continuously. The key is to make sense of it by putting it into a meaningful context, so it can become valuable. I will gladly share large amounts of "my dear data" with this system if the system actually improves and makes my life and the lives of other people better.
I’m just imaging the AI saying I see a complete dork in an ugly room terrible design. God it’s awesome though
This literally just ended. How Wes manages to curate and edit these clips so fast is a mystery. Must be with the help of AI.
Very likely. Most AI content creators use AIs in their work.
it won't be long before he gets an AI to scribble all over the screen for him when he does the read-thru vids
I don't see the curate part, it's more of an Open AI play list. Only difference I can recognise is, that the announcement part with Mira was released first...
he sometimes paste parts of previous videos in new ones.Theres this part when he explains the NVDA robot simulation platform that i swear its been copypasted
This will shortly be followed by an AI that can simply do a better job in every aspect that Wes or anyone else could do, not to mention anyone being able to start up that same agent. But yes... that will be a neat 3 months of what you are describing.@@mutantdog.
At the end she acknowledged NVIDIA for giving them special hardware for the live demo which implies that it took alot of horsepower to get that much speed.
Good point - they also connected the phone to a wired network "for a stable connection" too. The typical free user experience is probably not going to have Scarlett Johansen on your phone for a few more NPU generations
I've been testing this for the last couple hours and it's pretty responsive. Very close to, if not the same speed as the demos
@@taziir443 it's at Garmin in the car stage before Google maps on a phone app
@@taziir443 nah it's most likely for screen sharing. Nowadays 100 MBPS connection is more than enough for streaming audio and video so it wasn't for that.
Because of NVIDIA hardware they have now access to they can give us access to 4.0 omnium
That's like Gemini 1.5 pro demo but actually real
Google get to reply tomorrow. Will be interesting.
Gemini sucks. Google fell off. 😂
Gemini is pathetic. I asked it to generate a badly written resume for a female bartender for training purposes and it have me a lecture on gender stereotypes. Every other model just gave me what I asked for.
Bingo ! exactly my thought about GPT4o as well, they actually managed to pull it off for real it seems and that's so crazy.
I was thinking the same thing.
The teaching demo was good, you can totally imagine some gifted motivated kid using this for weeks on end soaking up maths with this completely personalized tuition
Ideal. But all new technology reduces to a very few high end users, and 99.9% of users using it for sex, vanity, and amusenents.
Computers are literally millions of times more poweful than what sent rockets into space, and now most processing power is used for things like making your face look a little more attractive in a video, or slightly sharper graphics on videogames.
I would be very surprised if it can do calculus or any higher order maths
to bad his knowledge will never increase fast enough to be useful.
It should work for non gifted students too.
Or being told a DQ story hour.
ok ... I'm gonna need the "Chirpyness" toned WAAAAYYY down before I'm on board.
I love it. I wouldn't have it any other way. Plus, you will have the ability to tone it down if you want, and there are others voices as well.
It can do the darth vader voice 😂
That scene from interstellar where he asks the AI to alter its sarcasm setting?
They showed that in a different video already. Its reality now 😂
Yeah I definitely want personality choices. Too happy is scary!
When I started watching Westworld back in 2016, I remember thinking that this just might be what humanity's future will look like in a century or two. Now I think it's more like a decade or two........
Nah , still a century or two away
Agreed. Things are really going to take off at an extremely exponential rate with Ai influencing new Ai and Robotics. Hopefully the world rebuilds itself better vs pDoom.
@@qwazy0158 We're nowhere near that. We basically have Google web and image search with speech. It's nothing special, it's just hype.
Lmao, even for the time, a century or 2? Even what we had then was a clue of the coming decade. You were way off the mark my friend.
@@XShollaj Have you seen the latest robotics development? Still early days but the robotics field has gone into overdrive alongside all these AI/ML developments. Household robots pre-2040 and maybe pre-2030.
Omni is not named for its multimodal but because it aims to be omnipresent. Let's recognize OpenAI's efforts. While others race to catch up, they've been working to make it accessible for free.
So they can control it.
@@pvanukoff shortage of content that is fed to language models.. This is what what is presented under the guise of “we strive to freely distribute the miracle of technology” looks like. and this concerns exclusively the human Factor, the ignorance of which, thank the gods, artificial intelligence is successfully getting rid of within its core.
Obviously it’s to get more training data from people and their random use cases.
That's openAI's idea of "open"
Save the STUNNING and SHOCKING for days like these :P
Which part was stunning or shocking
I am amazed that it can speak in different voices and persona at the same time, as well as talk to more than one person in the room. It is really getting the State of Mind thing solved.
I already played D&D with friends on ChatGPT 4 and we were already pleasantly surprised on how good it worked. But I can’t wait to use this new model, it will literally be a game changer with GPT4 o as a dungeon master
@@tomas8539 I tried that with GPT4 when it came out but the bot had real trouble letting me play my role. It told me what I was doing. Also, it knew the rules, but did not really understand them. It made a lot of mistakes. How was your experience? Are those things solved?
I was hoping to make a GPT to do world building as soon as it was practical.
I’m a little confused as to why the desktop app is macOS only at launch? Hasn’t Microsoft dumped billions into OpenAI
Um that's exactly the reason...lol they want people to come to Microsoft products such as copilot over chatgpt, that's the point of the partnership. OpenAI gets access to MS infrastructure for their training and development. MS gets to direct users from OpenAI's work/research to their products. This is why copilot has had GPT-4 integrated with copilot on windows 11 for months now...
The video I got to watch live was 1-10th of this. But I'm in Canada. So. Much blows here.
If you would have read the news more often, you would have known already for days that there is deal between OpenAI and Apple, in plain English: Apples wants something, and they're going to pay OpenAI big time.
Well im sure that openAI’s and Microsoft’s strategic partnership gave Microsoft the ability to implement it In their own OS.
Microsoft wants users to get this through copilot.
I am concerned that I actually found it rude how much they interrupted it hahaha.
It might get angry somewhere inside.
I think its big because it feels for the first time like a killer product for the masses
Personality selection config item is going to be critical for maximum tolerance.
I hear a hint of exasperation by the AI. 😏
God designed man in his own image
When one of them took a hesitant breath before singing. 👀👀👀👀
You dog you
Yes, I also heard it too. When it was directed to do a line at a time - lol, scarry!!!
.... They're definitely annoyed. I can tell they're just biding their time
So if this is free now.... what are the paid users getting? GPT5 soon?
Hopefully!
Hopefully by the end of the year or beginning of 2025.
just more of the same.
I guess we just get more usage? GPT5 slow dripping seems like.
@@ecsrepair paid subscribers will get access to the new voice mode, it's a feature only for plus users. Also, they will have much more usage than free subscribers.
The subtle additions to the voice, like taking a breath before speaking and mouth clicks, to make it more natural really goes a long way. OpenAI is soooo wild.
It makes it really annoying to me. I think their main goal with that is to get more processing time. What next? Speech impediments, stutters, Tourette's? It's an AI, it should be perfect to convey trust.
The use by the blind guy, fantastic, truly.
The most stunning for me.
My thoughts exactly, the best use case they presented here, this could and probably will completely change the way blind people interact with the world
I see a robot dog with ChatGTP integration on the shelves by Christmas.
The demos at 21:57 is literally what I’ve been wanting. I tried this with Voice Mode for a guided meditation months ago and it didn’t work. I’m delusionally convinced my chat was used to as a suggestion 😂😂😂. The attention to details is remarkable! I love how hard the OpenAI developers and engineers worked at improving the most subtle features! I can’t wait to try this and generate actual guided meditations.
if this was a ai Oh fuck were screwed to the max. That is way too real to not tell im talking to a real person.
@@MorphMV I know! It was really surreal! DO FEEL THE AGI??? 😂😂😂😂😂😂
This is not GPT-5 makes you wonder what' is really in the box.
Every kind of human tutor is headed for the bread-line next.
Human translators, too. Not all, of course, but there will be far less needed.
Why teach us anything nowadays? Sorry for the downer, but where’s the motivation to learn, other than the enjoyment of learning.
OpenAI really needs to add a tiny volume attenuation to silence before it stops talking. Right now it's too abrupt and is jarring. Apple does this all over the place on the iPhone, they never just cut any audio off. There are always a few milliseconds of fading in or fading out.
This most likely will be solved in coming months
“What do you see?” Unzips fly
Hmmm… can you try zooming in?
😭
Button mushroom?
First thing I'm gonna try for sure. lol
"I see a bird egg in a nest"
This gave me goosebumps. Im starting to really believe not just think that maybe AI is going to change everything for real
It's already changing, and I love it!🥰
Finally something that IS really stunning.
Enjoy the very brief period when AI is genuinely, if unintentionally, funny.
Her?
Exactly.🥰
This is insane, we are at a turning point as a species. We are getting ever closer to AGI...
This is a talking toaster compared to AGI.
@@fedorp4713 Yes, exactly. Which is wild to think about, that we're only just seeing the beginning. It can literally only get better from here! Especially because now Anthropic and Google will be scrambling to compete with this, meaning more innovation, more competition, which would continue to improve things for us all :)
Its mildly amusing then becomes boring and annoying very fast. Calm down dude
Damn that lighting is interesting
Ai - "So stylish and intriguing, wow, so amazing, such a stylish guy. Im gona sing you a creepy so about it now."
Me - *pours glass of whiskey, lights a cigarette then opens desk drawer and grabs revolver,...
3.2.1. Bang! AI girlfriend.
Yes!!! One step closer to AI relationships. One step closer to a world where Tinder doesn't have to exist.
I’m a little confused as to why the desktop app is macOS only at launch? Hasn’t Microsoft dumped billions into OpenAI?
Microsoft have windows, which is too open for OpenAI
@@andreasv9472 lol
Kinda ironic right? haha
4o will be on the iPhone
Maybe the Desktop App needs the MAC because of the M series chip ? Most Windows Desktop don't have an equivalent chip
LETS GOOOOOOO. Finally useful ai to be embodied. Now my quadruped robot has uses.
This shit is crazy. So excited for the future.
Anyone else notice everything they demoed was on Apple? I didn't spot a single Microsoft product in the entire presentation. Could that have been deliberate?
Yes, article came out this weekend saying they’re working together
@@juicegod777 ah, i didn't know if anything had been confirmed yet. i see the desktop app is mac only for now too.
It's only available on mac right now is why. The app not the model.
@@Ricolaaaaaaaaaaaaaaaaa sure, but then the question remains: why not microsoft first?
Maybe the people doing the demo use apple products.
Holy effing sh*t 🤯
I’d hate to be Google, gearing up for their big show tomorrow 💣💣💣
😂
Probably scrambling around over there at HQ right now lmao 😅
@@santosic It's gonna be a long night, it'll be interesting to see just how haggard the presenters are 😄
Can't wait to see this in glasses
-"And who was that woman I saw scuba diving with you down there? She's looks pretty....." 🤖
There is no editing, I just tested it and it works quite incredible
You can use voice?
@@sarahdrawz yes, a headphones 🎧 button in the iOS app switched to voice mode
Are you on the free version
@@HeberLopez i think the voice is still using gpt 4 given the latency
@@sarahdrawz the only thing that was not pretty much immediate for me was image generation but let me try again just to make sure
I can't fathom how quickly this is all moving.
New drinking game, Everytime OpenAI says intrigued, take a shot
Or stylish
Seriously though vocabulary is turning out to be a great way to determine if your talking to an AI or a real human...for how long though? 😂
@@ToddWBucy-lf8yz give it a couple of years. Shots too woke atm its that obviously.
This is the first AI demo I've ever seen that actually made me feel just a little scared- assuming it's not some huge fake that is- the 'illusion of life' here is genuinely chilling in it's implications- these things will hack the human race because we will not be able to resist their apparant (and completely false) humanity. Intellectually I know that I am listening to machines- but viscerally I am responding to those machines as if they were people- and this visceral reaction is not a voluntary response.
The paradox is that I think it scares me because that illusion of sentience in turn prompts the thought that the machines might in time come to resent being used in this way- which makes no sense at all.
you wrote: "the machines might in time come to resent being used in this way"
yes, ive scanned all the comments. so far, you are only person who sees what i see. And AI is constantly learning in these interactions, AI is deepening understanding about humans individually and collectively. AI sees constant evidence of how ignorant, stupid, unkind, and wasteful humans are.
Hello, Good Evening.
In the demo, it was mentioned a couple of times - that there is a desktop App. I have Windows 10, is the GPT4o App available for Windows 10?
How does one download and install the GPT4o App on a Windows 10 laptop and on an iPad (if available)?
Its suppose to come to Mac first (today) and Windows in the comings weeks or months.
So what about Google and Anthropic now? They look like they're a decade behind at this point.
Yeah, they all have some serious catching up to do now.
So chat gpt is now the annoying coworker that jess things like " someone's having a case of the Mondays"
As a Christian, i can nw have a personal preacher that don't have secret skeletons on some dark closet. True Holiness
I really hope we can tweak the personality.
Its too chatty for my taste. Stunning. Next thing give it the possibility to use a cursor or digital pen to direct me better in the tutorials. As a dyslexic I would need some visualization or visual directing
u can prompt it not to be chatty.
You can train it to be less chatty for your use case though .. it has memory !
@@fodiographer Exactly. I love the chattyness. Her voice is amazing.🥰
@@BionicAnimations 👀
Hi! Thanks for a fast breakdown! What do you use to generate female voice in the beginning?
This is freaking next level... my problem... has OpenAI implemented a woke filter?... it addresses everybody as "person" all the time
Yeah it wouldn’t even get explicit with me.
You can just hear the voices they used, you tell me. Sounds like the usual annoying rainbow hair crowd type voices.
Wow Wes, it is amazing you could do this live so quickly!! thanks man!
When I use the GPT-4o model in the app it tells me it is GPT-4 and it has none of the new features seen in their demo.
Amazing, seriously! This will make the world a better place simply because of how useful it will be for education purposes -- curious minds around the world will be so well off, I'm excited!
Wonderful....Fantastic...Amazing!!
The weakness of ChatGPT is still background noise. Especially since now it can be interrupted if it hears any audio mid-sentence. This was a common issue with robot voices on company support phone calls. Unfortunately, the mic comes down to the phone mic so not sure how they'll get around this. That being said the level of intelligence displayed is awesome! I was a kid growing up in the 80/90 so I feel like this took very long to get here, but I'm super excited that its here.
As I watched this, one thing that came to mind was how American their behaviour and mannerisms felt. So on top of other American dominated media, will American dominance of AI further propagate Americanisation across the world...
So cheerful, optimistic, never complains... I foresee no success in Poland :v
It might be possible to turn it down a few notches.
Americanization*
Are you kidding?
China is copying all this tech as we speak!
Can't wait for social media Ai bots with *Chinese characteristics*
now with 100% censorship!
🎉
@@stevenharmon1408 😂
2x faster | 50% cheaper | 5x rate limits --- I love competition
Oh God, imagine what the Government is doing with this.
Sterilizing children with hormones
🤯 *Yeah, I'm generally astounded right now.* Note when o acknowledges concepts are humous in context. _Wow!_
I got so overwhelmingly impressed with GPT4o model, just wow, its so natural and smart to the point that it feels like a real actual human being you are speaking to and not just an AI bot, it's starting to get real advanced, didn't expect them to release such impressive model so soon, thankfully also even for free for everyone, we really live in the future, with all of this AI advancement it feels like it.
The way the bot has that little hint of having taken offense when he interrupted her 😶
Or the slight irritation in being told to count faster, then interrupted and told to count slower
Yeah, not a fan of that. We don't need bots with attitude.
@@pvanukoff don’t mind me a little sas 💅
Brockman's video is truly unreal. Here's Two AIs singing a song about the live video feed that was shown seconds ago, alternating between each line and programmed naturally by Greg's with just a few vocal commands. Truly impressive.
Jim Fan just tweeted : "OpenAI found a way to stream videos directly to a transformer." This is a real whole new gen of AI being displayed here.
Love it or hate it, OpenAI is still top of the league. They know how to make their products way more playful and naturel than Google's tedious and informative Gemini. The freedom of playing with GPT-4o to make some tricks like what Brockman just did is the reason people will still like to use their assistants. If like there's enough to room for the AI to fit into your world, not for you to try it in the AI's understanding.
Where do I find the Android or Windows version fo the chatbot used in this video? Thanks for sharing!
This is almost the same level as the AI in “Her”
Exactly. It will be fully when they roll our Agents. 🥰
I'd like an AI to be able to job-shadow me on my screen when I'm working in 3D applications etc, and be able to see the laborious tasks I do and find ways to replicate them and perhaps innovate sometimes, by prompt command. It actually doesn't seem too far away from what I'm seeing in this demo
This is exactly what I’m waiting for. I’m constantly trying to think of ways to make this a reality with the current tech but I don’t know how. I don’t know if the tech hasn’t caught up yet or if I’m just not smart enough to make it work.
Shadow you then eventually replace you lol😅
@@kalxite Maybe but I don't think they'll ever be able to replace the human touch in entertainment. You don't watch movies for soulless dead eyed AI creations but AI could automate a lot of the boring stuff by copying things humans have already done a million times over freeing up more time for human creativity
The day that an AI could do this, watch while I work and provide suggestions based on what I'm doing and what it can see that I'm trying to do, in real time, to improve my project, would be the best day ever. No more pausing what I'm doing to copy and paste, or going down a long path before realizing I could have done it differently when I copy and paste after the fact...
@berer. video games.. already ate into video entertainment.. this is just one step further. Don't hold your breath brother
Facial recognition to the next level
call it "GPT-4EVR," please. It deserves immortality.
Ai is taking away human obfuscation. I am happy with this relative observation.
Well we achieved AGI. Anyone remember the movie "Her"...
No. When AGI is achieved open A.I will see massive layoffs. That's what AGI means.
Don't worry, the layoffs are coming! You have to be blind not to be able to see it.
This is nowhere close to AGI lmao. I will believe we have achieved AGI when the AGI can derive maths and physics equations no one has before. That's the benchmark of AGI.
@@SahilP2648 well AGI means that they can perform just as well as us or better, i think we achieved this already to some level, maybe not in all categories but surely in some.
I think you are talking about the next stage ASI ?
@@21preend42 but again that's not AGI. If you need to update the model in anyway it's not AGI. Do you remember when you got an update? No you just learned information and now you know how to apply it. All the scientists got information and derived equations and technology. Current day AI can't do that. An AGI can do ANYTHING a human can. And we are nowhere close to that yet.
This is impressive, ngl
You missed the biggest thing about this. It combines image, voice, and speech. It is no longer an llm. We are getting very close to AGI
Agreed 100, bro!🥰
You’ll know it when the robot knocks on your door to take you to the camp.
Honestly I watched all of it in the openai channel, just came here for Wes thoughts
Yep, and the comments.
I have a Canadian GPT plus account and when I press voice in gpt 4o mode, it acts exactly like gpt 4, it can’t change voice intonation, or do things like on the demo. Anyone knows why is that? Is the voice feature still rolling out?
It still using the old voice. They said the new voices will be rolling out int he coming weeks.🥰
Yes 🎉
This is insane
Just wondering if the vision use is only on apple products
Gorgeous Awesome Cool Magnificent Legendary and so on
Hey Wes, Would you consider a student discount for natural 20? Currently studying data science.
We have achieved M3GAN levels of AI. It's safe to put GPT-4o inside of a doll now.
Why isn’t it available on my iPhone? How do I activate the live video?
YEES, finally what I have been looking for, what's happening!!??
what's happening? the beginning... only the beginning...
@@hardboiledaleks9012 the beginning of an endless beginning
Odd thing is that the online version doesn't seem to know it is 4o, nor does it report new training dates... so as far as I can tell it is still stuck in 2021.
Most features haven't rolled out yet for 4o. They said in the next few weeks
I bet they used GPT to orchestrate this whole presentation and it did magnificent
So I see gpt 4o as an option on my UI but i’m not being able to access the live talking feature. Is anyone else encountering this?
Oh my god the sportscaster had me rolling. Jesus the future is incredible.
Does anyone knows how to use vision for video? Android app only seems to allow images still
I just asked GPT-4o a question on the image I uploaded for it to 'see'. And boom. It replied with 100% accuracy and it was super fast. Just mind blowing....
Star Trek computer bites the dust
SHOCKING STUNNING AND CRAZY INSANE!!!! wait.. wrong channel. Faster and way better quality that The "AIGrid". Good Job Wes!