Say hello to GPT-4o, our new flagship model which can reason across audio, vision, and text in real time. Learn more here: www.openai.com/index/hello-gp...
to be fair, that's what Siri felt like for the better part of the last decade. Except not a proper, regular cassette player, and more the kind that eats up and demolishes the tape in the process Siri has been hot trash for a good while now
@@holyclasher a while ago google demo-ed gemini ai, which they advertised as a real time feedback ai. In the demo they played a game of rock paper scissors with the ai. However, it turned out that the demo was scripted and they received a lot of backlash from it.
How did it happen that they both used scissors for two times in a row? Did they plan this? If so how would they know the LLM would choose rock, paper, scissors game?
im sure theyll eventually introduce an update so it can remember you or other people, make your own ai actually feel personal. I wouldnt be against it as long as I can choose to not share the data it collects
this is true agi man. im blown away. the infantile steps of agi in our hands just with a subscription of 22 dollars per month. id honestly pay even 100
The changes are very deep especially in the personnality. Tbh, it's how I imagined GPT5. So if it is only a GPT4 upgrade, then what will GPT5 be able to do ? What a world
She's smart, funny, charismatic, creative, entertaining, great to talk to, has all the knowledge of the world, basically everything a guy could ask for and she was just born today.
@@italiangentleman1501 imagine ai bot looking like similar to an actual girl with body parts and warm soft holes and you can customize anything about her.
The key difference in this model, is that is not generating text, and then using a different text to speech model for voicing. It is natively multimodal. It takes audio, image (not sure about video), or text data, and spits out audio, image, or text data. This allows it to be trained on all of those modalities, and to be finetuned to respond in particular ways. It was trained on probably millions of hours of audio speech, and as such picked up on the general expressiveness of how humans talk.
This is truly amazing… But watching this I can't help but think about how truly hard it’s ever been up until this point to train a computer to simply recognise any of the hand gestures… let alone give vocal feedback in real time - and soon everyone will have access to this… definitely historical moment! ✊✌🖐
@@bblunderis being biological an advantage though? Living creatures are so expensive to maintain, they get sick, they die, they cannot make a perfect copy of themselves, they unconditionally believe in things like god... maybe being biological isn't that great after all
Yup, not only that but it also remember what round they’re on, remembered that the first two turns were a tie, and when the third round started it said “third times the charm”. This is truly getting crazy. I think it’s really cool but I am genuinely getting terrified now. This capability could easily replace customer service jobs, and even replace referees for the NFL, NBA, NHL, etc. This technology truly is going to take away so many jobs and I don’t think people grasp how serious this is getting.
@@Coryo61827 We will find a way to navigate it. People felt the exact same when the internet / google came out. It’s a human instinct to fear something new, it’s a safety mechanism to protect our own. Yes people lose jobs but it opens doors to new jobs and opportunities. Who’s to say ai doesn’t solve all humans problems? It was made by humans after all.
@@Coryo61827 especially with the speed of improvement, they were able to get from pretty good text fo fully multi modal in a little more than a year another 2 years from now we are cooked. We need universal basic income and we need it SOON.
Its insane that these only have a few thousand views even if its only been an hour,this is unbelievable, and its only going to get better, im sure at some point we'll look back at this and think that this is primitive. And then AGI will be made and thats basically be the end of it. (Im not saying It will kill humans or anything it will just probably eventually do everything that humans used to,including improving AI)
I see this as just another step in the evolution of humanity, it’s just not a biological one anymore. By creating an AI in our image we can pretty much become immortal and spread across the entire universe (assuming it‘s physically possible), something we could never have achieved in our current, biological form.
@@cholobok Yeah,I guess I do also think that,at the point that AI/AGI makes humans pretty much useless, I dont really mind humanity (or the biological form of humanity) dying out
@@EchoMountain47 Yeah ,heard it, unbelievable. The voice doesn't even sound exactly like her and didn't they say that the actress for the voice wasn't scarlett johansson? It's kinda wild how can someone strike someone for copyright just because the voice sounds similar, even though they didn't use samples of Scarlett Johansson's legit voice. What a crybaby
when this is public launched, I hope to see whether this demo of rockpaperscissors can be completed by regular people and if it can be any regular drinking game...
When are we going to have this version, because currently eventhough I selected the latest model(gpt4o) it is not like in the video. I think this still would be launched right? Thanks,
Welp, there goes customer service jobs. I don’t think people grasp how much trouble we’re going to be in in 10-20 years. There won’t be many jobs left that still need to be done by people. And the jobs it does replace, those people will migrate to other professions, further diluting the pool of candidates, making the chances of you landing a job drop dramatically.
I wonder what would have happened if one of them protested and said 'No I clearly have rock'. Imo AI is not too determined and will always agree with you if you think it's wrong
OpenAI team, if somehow you’re reading this, please at least reenable the legacy Sky voice for earlier versions of ChatGPT, as it’s safe to say it sounded nothing like Scarlett Johansson. I can understand the controversy with 4o, but it’s actually really disruptive when you’ve gotten used to a voice for eight months and it is taken away suddenly. Please please please bring it back
@@skyplayx3558 Yeah, it's multimodal. I'm just wondering how it got to learn speaker segmentation. Is it one of those things the model suddenly learns after much training? Or maybe with their sentiment analysis they also got the power of giving each voice an id
I did not expect Rock, Paper, Scissors gameshow host to be the first job lost to AI.
🤣
lol
dang I just started today too
😂😂
well there goes my degree... fuck this shit
Makes Siri look like a cassette player.
Radio even
lmao
to be fair, that's what Siri felt like for the better part of the last decade. Except not a proper, regular cassette player, and more the kind that eats up and demolishes the tape in the process
Siri has been hot trash for a good while now
Siri is absolute garbage
Not for long. Apple has been doing some acquisitions in the background. Future Siri will have all these and more i think 😏
this is definitely a shot at Google 💀💀💀
im out of the loop can anyone explain
@@holyclasher a while ago google demo-ed gemini ai, which they advertised as a real time feedback ai. In the demo they played a game of rock paper scissors with the ai. However, it turned out that the demo was scripted and they received a lot of backlash from it.
That would be a bit like the USA "taking a shot at" Easter Island. OpenAI have always dominated Google in the Gen AI product space.
@@autumn.variation wow just watched a video on it thats funny they just fed it images and gave hints something gpt could already do
@@autumn.variation can you tell how to find that video pls
No Gemini AI were harmed in the making of this video 😁
😂
Gemini: We will make Rock, Paper, Scissor game even faster with IF, ELSE Statement 😂
gemini crys in the corner You'll see in a few years!
@@AshokKumar-mg1wx no... "Rock paper scissors seems to be too violent to be allowed in general public. It's racist to play Rock paper scissors"
Is this an attack at Google Gemini’s faking “rock paper scissors” demo debacle awhile back 😂
gemini destroyed, never had a chance
A clear shot indeed 😂
Gemini is sh*t
absolutely lol
How did it happen that they both used scissors for two times in a row? Did they plan this? If so how would they know the LLM would choose rock, paper, scissors game?
I like that gpt-4 O knows who is who. I also wonder up to how many people it can identify by voice, picture and name and how it recognizes who is who.
yeh actually a really interesting question
im sure theyll eventually introduce an update so it can remember you or other people, make your own ai actually feel personal. I wouldnt be against it as long as I can choose to not share the data it collects
We can start using it in some weeks!
@@coreywalker2763 this update has memory between sessions, so wouldnt it be able to do that already?
@@coreywalker2763 theres no way you would know whether they honor the agreement. sam altman is literally trying to scan everyone's eyeballs
my future as a dev is as bright as the dark side of the moon 😭😭😭
It’s gettin crazy bro 🥹
Your future gets bright every 29.5 days 🌝
This thing is gonna be a mini-product manager in a years time
It's like we're all suddenly breathing something that's no longer air.
WOW. Again, it's the speed that's the most impressive overall.
Indeed, its amazing
the expressivity and comprehension too
this is true agi man. im blown away. the infantile steps of agi in our hands just with a subscription of 22 dollars per month. id honestly pay even 100
The sports announcer voice put this over the top for me. Well played ChatGPT4o
The changes are very deep especially in the personnality. Tbh, it's how I imagined GPT5. So if it is only a GPT4 upgrade, then what will GPT5 be able to do ? What a world
Not a lot. All of the internet won't be enough to teach gpt futher
@@tenzingyatso5096 replying to you just so i can find this comment in the future to see how wrong you were
@@tastymuffinmmSame 😂
@@tastymuffinmmsame
@@tastymuffinmm check out numberphile video about it
She's smart, funny, charismatic, creative, entertaining, great to talk to, has all the knowledge of the world, basically everything a guy could ask for and she was just born today.
bruh, are you really grooming an AI? 💀💀
Why don't you take a seat
@@italiangentleman1501 imagine ai bot looking like similar to an actual girl with body parts and warm soft holes and you can customize anything about her.
@@winnershandbook1069 calm down blud💀💀💀
Outrageous
Mine with Scalett Johansson's voice, please.
it's kinda, when I first used the AI voice in gpt3.5 I thought it was scarlett, and I think it should sound familiar, gives a "Her" vibe to it
Do you mean with her voice?
@@valbermits a movie
@@user-hi2dh4ug4t wait gpt3.5 has a voice?
I’m sure there will be a vibrant marketplace in the future where you can purchase any celebrity voice you want.
HOW DID YOU MAKE HER SO EXPRESSIVE
Update is coming out in a couple weeks
She has access to tik Tok and insta
The key difference in this model, is that is not generating text, and then using a different text to speech model for voicing. It is natively multimodal. It takes audio, image (not sure about video), or text data, and spits out audio, image, or text data. This allows it to be trained on all of those modalities, and to be finetuned to respond in particular ways. It was trained on probably millions of hours of audio speech, and as such picked up on the general expressiveness of how humans talk.
@@gundulfguy2179 Its impressive at the least!
@@davii_ms thats for sure
This is truly amazing… But watching this I can't help but think about how truly hard it’s ever been up until this point to train a computer to simply recognise any of the hand gestures… let alone give vocal feedback in real time - and soon everyone will have access to this… definitely historical moment! ✊✌🖐
And this is not even its final form
@valberm it's scratching the surface in its capabilities, GPT5 is going to be insane
Great point. Recognizing hand gestures, facial expressions itself was a bit deal. Now it's interactive wow
Bigger than original iPhone release
keep doing scissors until it knows it's been tricked
Good idea!
OpenAI shocked the world today - great product release! 😀😁
OpenAI shocked the world today - great product release!
The way they cut her off☠️
Incredible that we went from living in caves to this
What is even more incredible that we are biological, and any smart AI will be not unless we really figure out how to play the god
@@bblunderis being biological an advantage though? Living creatures are so expensive to maintain, they get sick, they die, they cannot make a perfect copy of themselves, they unconditionally believe in things like god... maybe being biological isn't that great after all
so.. expressive.. these engineers must be having so much fun building this
Yeah must be fun building something that will take away your job soon, im sure they having a blast!
Incredible it visually remembers which name belongs to who when it announces the winner
Yup, not only that but it also remember what round they’re on, remembered that the first two turns were a tie, and when the third round started it said “third times the charm”. This is truly getting crazy. I think it’s really cool but I am genuinely getting terrified now. This capability could easily replace customer service jobs, and even replace referees for the NFL, NBA, NHL, etc. This technology truly is going to take away so many jobs and I don’t think people grasp how serious this is getting.
@@Coryo61827 We will find a way to navigate it. People felt the exact same when the internet / google came out. It’s a human instinct to fear something new, it’s a safety mechanism to protect our own. Yes people lose jobs but it opens doors to new jobs and opportunities. Who’s to say ai doesn’t solve all humans problems? It was made by humans after all.
@@Coryo61827 especially with the speed of improvement, they were able to get from pretty good text fo fully multi modal in a little more than a year another 2 years from now we are cooked. We need universal basic income and we need it SOON.
Her는 못하던 영상통화가 되다니 초대박이네요. Kudos!
Technology is moving scarily fast 😢
Especially the AI technology
@@bblunder a real shocker.
Its insane that these only have a few thousand views even if its only been an hour,this is unbelievable, and its only going to get better, im sure at some point we'll look back at this and think that this is primitive. And then AGI will be made and thats basically be the end of it. (Im not saying It will kill humans or anything it will just probably eventually do everything that humans used to,including improving AI)
hopefully its utopia from then on...
I see this as just another step in the evolution of humanity, it’s just not a biological one anymore. By creating an AI in our image we can pretty much become immortal and spread across the entire universe (assuming it‘s physically possible), something we could never have achieved in our current, biological form.
@@cholobok Yeah,I guess I do also think that,at the point that AI/AGI makes humans pretty much useless, I dont really mind humanity (or the biological form of humanity) dying out
Finally! This is the update I've been waiting for!
such good ideas !
And here I cant even get Siri to do a voice call to the right person
I hate it when they interrupt the AI
Me too. Lonely people will become a pain in the ass if they get used to talking to a pushover all day long.
Awesome!
Now it recognizes who's who. Impressive!
Imagine implementing this in video games, we can start talking with NPC's in real time lol
The next generation of gpt is going to be even more impressive. With videos and photos of people being used for gpt enhancement
I'm hoping the next version doesn't require telling GPT to stop talking so much.
Yeah it's too ott atm. Kind of annoying, I want a more chill minimalist one
Very amazing omg
I imagine in the future people will constantly interrupt each other mid sentence, because they are so used to it when working with AI
OpenAI always on top :)
just incredible
This is going to be Jarvis turning to Ultron story becoming real real soon 😅😂😂😂
The ultimate showdown of the century? She's giving us hints, don't trust her, she's gonna end us all! 🤣🤣🤣
It really improves the game!
I wonder what kind of database you guys have. It must be so crazy :0
This is something i wanted.
Wow, so innovative.
So good billions were put into this
Totally worth it
I detect sarcasm and let met tell you; that is a mistake. AI is changing the world by the minute these days.
they feel so rude wen they cut the ai off I know its an ai but i can't get over it
It's probably a time constraint thing. They probably have to keep the video under two minutes as a demo
So cool
Hahahaha :D that was funny! and cool technology! good job guyz!
OpenAI just showed Gemini how it's done.
that AI genuinely has a beautiful voice tho
So true 👍
Yep
@@luthenrael4523 It is designed to appeal to the widest possible audience
Yeah, not anymore. Scarlett Johansson whined so OpenAI took down the voice. I doubt they’ll bring it back given all the controversy
@@EchoMountain47 Yeah ,heard it, unbelievable. The voice doesn't even sound exactly like her and didn't they say that the actress for the voice wasn't scarlett johansson? It's kinda wild how can someone strike someone for copyright just because the voice sounds similar, even though they didn't use samples of Scarlett Johansson's legit voice. What a crybaby
The important point is that she can distinguish who is Mi or Alex
Would it have understood if the paper was held horizontally instead of vertically?
If this is how we gonna play rock paper scissors then i am gonna hold tournaments
Super!
when this is public launched, I hope to see whether this demo of
rockpaperscissors can be completed by regular people
and if it can be any regular drinking game...
Lovely demo! You have to fix that audio though
The scary part is that the AI sounds more human then the engineers in these.
I am not able to use this continuous video chatting in my android is this available only for iOS users right now ?
Is the rate of 40 rounds per 3 hours is enough to finish this conversion for normal plus users?
A little shot at Gemini there lol nice
This is dream assistant for all phones
Amazing
Can't wait to see her replacing Ludwig on Mogul Moves.
When are we going to have this version, because currently eventhough I selected the latest model(gpt4o) it is not like in the video. I think this still would be launched right? Thanks,
ChatGPT Plus subscribers will get these additional features in a few weeks, and the general public will get them sometime after that
Wow. It's pretty good 😅
Open ai, pulling a prank on us 😂
Imagine what they're sitting on with gpt5 if this is just gpt4 level intelligence
Amazing 😂
incredible kk
When will this be available for public
few weeks
And is FREE
This makes chatgpt look like CD disks in the age pen drives
Girl this is ChatGPT
@@eemoot I know dog. U get the context 😜
@@maheshBasavaraju lol 🥂🤪😝🥂
Google in shambles right now
Is it only iPhone version have camera? I tried android not support
Her voice is attractive
The count down for rock paper scissors is 'row sham bow'
is all this visual chatting only working on iPhone? cause on Android its not available
The update hasn’t fully released but it will in the coming weeks
woah! that was good!
What will you do if the AI says, “don’t interrupt me when I’m talking”
What brand Alex wearing?
No one noticed the Freudian slip @0:08 That's absolutely hillarious!!
From what I understand, this camera capability is not yet available for the general public, right?
how can I interrupt her? do I need to pay GPT Plus? cause either the camera nor the interrupting action works for me
Welp, there goes customer service jobs. I don’t think people grasp how much trouble we’re going to be in in 10-20 years. There won’t be many jobs left that still need to be done by people. And the jobs it does replace, those people will migrate to other professions, further diluting the pool of candidates, making the chances of you landing a job drop dramatically.
WoW!
Nice
A real shocker..
I wonder what would have happened if one of them protested and said 'No I clearly have rock'. Imo AI is not too determined and will always agree with you if you think it's wrong
God dammit this is cool
I thought GPT-4o would do a slow-motion replay to judge if both players extended their hands at the same time.
Rock, paper scissors really??? No fifa??? 😂
paper hand gesture is too carefully presented
Why there are using iphone???
Is it real😮😮😮😅.
It's great.
OpenAI team, if somehow you’re reading this, please at least reenable the legacy Sky voice for earlier versions of ChatGPT, as it’s safe to say it sounded nothing like Scarlett Johansson. I can understand the controversy with 4o, but it’s actually really disruptive when you’ve gotten used to a voice for eight months and it is taken away suddenly. Please please please bring it back
Hope it can host LoL world finals
So is this available now or what?
in a few weeks.
In few weeks
It remembered their names 1 minute later
Wait wait, how did it knew who was who?? Does it uses video and understanding of lip movements? Did it guessed by gender?
oh my sweet summer child
It's a multimodal AI. It doesnt do any of that, it just does a bunch of math and that somehow becomes GPT-4o
@@skyplayx3558 Yeah, it's multimodal. I'm just wondering how it got to learn speaker segmentation. Is it one of those things the model suddenly learns after much training? Or maybe with their sentiment analysis they also got the power of giving each voice an id
Can it beatbox??
🤯🎉🎉
crazy
The humanity is saved