ChatGPT Can Now Talk Like a Human [Latest Updates]
Vložit
- čas přidán 19. 05. 2024
- Go to brilliant.org/coldfusion for 20% off!
ChatGPT took over the world in 2022, but since then things had settled down... Then came GPT-4o, a human-line voice assistant done in a way which we haven't seen before. In this episode we'll see what it means for education, productivity, the future and much more.
Sources and show notes: docs.google.com/document/d/1t...
ColdFusion Podcast:
/ @throughtheweb
ColdFusion Music:
/ @coldfusionmusic
burnwater.bandcamp.com
Get my book:
bit.ly/NewThinkingbook
ColdFusion Socials:
/ discord
/ coldfusiontv
/ coldfusion_tv
/ coldfusiontv
Producer: Dagogo Altraide
Writers: Dagogo Altraide, Tawsif Akkas
Editors: Tanzim Uddin, Dagogo Altraide - Věda a technologie
![How We Became the Loneliest Generation [Documentary]](http://i.ytimg.com/vi/I19btmIBhx0/mqdefault.jpg)
![How We Became the Loneliest Generation [Documentary]](/img/tr.png)
A quick update for those who don't know: A few hours after this video was published, Scarlett Johansson (the actress who voiced the AI assistant in the movie "Her") released a statement blasting OpenAI for using a voice in ChatGPT (which they called "Sky") that is way too similar to hers.
You should definitely read the full statement but many online noticed the resemblance too between the two voices. OpenAI said they didn't use Scarlett's voice but has decided to pull back Sky from the app for now. All this while a few top member of the OpenAI safety team also resigned.
Not too sure what to make of all this from the outside but it does seem a bit messy to say the least...
it was a sleazy thing to do. sam altman seems like a sleazebag
Sky is available on the app as I have it at the moment.
Scarlette Johansson was approached by Open AI and she refused, They asked her another time before going public 2 days prior and she didnt have chance to respond and they used a voice that is said ot be "similar" to hers, Open AI is saying they used a voice actress so SJ lawyers are seeking evidence of this, which if they did use a voice actress that is the end of it, regardless of similartity she cant say nothing and it becomes a missed opportunity for royalites for her and fame, you could say a "Kodak moment" for those who know the history of the digital camera will get the reference and their slogan, but if they did use it without permission there will be some compensation no doubt.
No wonder I didn't see that voice in the speech settings recently.
It's not "WAY TOO SIMILAR", it's just the intonation which you can change anytime you want.
@@GabeHowardd its sleazbaggery and its been removed because he doesn't want to be sued
Born too late to explore the world, born too soon to explore the space, born just in time to see the rise of the machine.
😂
AI can nuke us to the outer space.
all your 3 wishes would have been granted in one go.
as you didn't specify in your prompt: alive
Say Hi, John Connor.
And possibly to be unalived by the very same Machines...
Space is already being "explored". Interstellar travel will NEVER b possible.
"TARS , turn down the giggling by 50%."
we will get there
You are a legend
There should be a setting for that
CASE: left the chat
You can already do that, the personality is crafted by prompts that you don't see
"hehe thats incredible rocky!
That's all down bad dudes need
That's a possible scenario. No great Terminator war, just anticlimactically get turned into AI's Pomeranians.
its a view into how these developers think people should sound, i mean how would they know, their nose stuck to a screen deep in some lines of code that can now be done by the same thing theyre working on. they program it to be that way.
@@romchompa6858 you know these models are trained on data, and not on the whims of the developers, right?
Just like one of the scenarios predicted by Ted Kaczynski. "Benevolent" elites keeping the rest of us dependent on machines for physical and emotional needs. Sooner than you think.
As someone said on another video about this ai, when it says "you look lonely, i can fix that". Then we've finally smashed right up against the scifi window and broken through.
I saw that comment, it's a small world
Customer service calls in like a year or so is gonna be wild.
The scam callers are on the way there. Which is less fun because you don't get to troll them... people like Pierogi, Pleasant Green, DEYO, Kitboga and IRLRosie won't have as much to do.
but imagine no wait, everybody gets personal help right off the bat, no menus. but then nobody will have jobs, no money, so no products and services to call and complain about
I would much rather this than someone in an Indian call centre that doesn't have a clue what I'm on about
AGI, Artificial General Intelligence will be run by A Guy from India.
keep your pants on
R.I.P the birth rate
We don't need birth rate. Humans are so outdated.
Why?
@@RiwenX I think you know.
That is the plan. First, AI reduces the human birth rate. Second, take over the whole WORLD!
It's the better of 2 evils. Better than starvation
Perfect timing for this release to help us forget about Ilya and the safety team all quitting lol
It may be related, could be they consider this model being so lifelike and charming a breach of ethics and safety. Plus with native vision and audio, this isn't just a LLM anymore. It has become something else.
@@Hodoss Yeah it is a LLM with speech and a visual model. Why are all the slow people trying to make it out to be more than it is. It is a random number generator with a database that is weighted and you can do database operations to adjust the weights. Stop trying to pretend it is magic, it only is if you are a very slow and ignorant person.
@@thomgizziz LLM means Large Language Model. GPT-4o is doing non-language things, so it's not just a LLM. Simple as that.
Less ranting and projecting your insecurities, more thinking, grandpa.
Cause they’re NOT going rogue 😂
"I'm sorry, Dave. I’m afraid I can’t do that"
I would be so much happier with that voice than this preschool teacher vibe.
@@willdarling1 Sounds more like a Valley Girl at the Beach vibe to me. Which is also cringe tho.
@@ArawnOfAnnwnLonely men love that voice!😂
😂😂😂
@@willdarling1 don't worry. Once people get used to it you will be allowed to use any voice you want.
"You won't be carrying a calculator with you all the time." - My math teacher in 1985.
I read that as “marrying” a calculator
😂😂👍
They were still saying that in 2005 too
Mine in fricking 2012. When I already did with my phone. Fricking tech iliterate dust collector.
my maths teachers 2005-2011.
We are passing the turing test with this one 🗣🗣🔥🔥
It was passed a while back, arguably. The basic one isn't that hard to pass.
@@Hodoss speak for the person taking the test... heh probably low expectations built into the test.
I think most current LLMs will pass the Turing test. But sadly these models are not human like intelligent. It seems the Turing test is not good to judge intelligence. These models are like an all knowing machine parrots, which has learned a billion phrases and uses them in the correct situation most of the time, but it does not actually understand a thing, if you elaborate deeper on what understanding it has about what it said and how it explains what it says it would get confused very fast and it's answers would become nonsense. The model can even confuse you, with itself... i.e. it doesn't recognize itself as a different entity. That's why it's so funny some people think these models are "dangerous"... "dangerous how..." they are just conversation completion machines, not AGI, nowhere close to AGI, they just confuse people like someone would get confused when he/she hears a parrot for the first time.
@@romchompa6858 It's always been a flawed test, Turing didn't mean it as a serious test if I remember correctly, he was just pondering.
But it's trickier than you'd think. There's games where you can try it, like "Human or not?"
czcams.com/video/0DIixhg2V_c/video.html
@@romchompa6858 Arguably it's a flawed test, Turing didn't mean it as a serious test, it was just a thought experiment.
It's trickier than you'd think. There's games that are essentially that test, like "Human or not?". Also videos of people playing it and struggling =D
spyware has become so sophisticated nowadays...
OpenAI - AGI= Artificial General intelligence
Google - AGI= Artificial General Idiocy
Amazon -AGI= A Guy in India.
Didn't you know that "AI" means "arrogance and ignorance"?
@@HansDunkelberg1 So, it stands for Donald Trump? ;-P
🤣🤣
@@igorschmidlapp6987 We found the Biden voter.
@@HansDunkelberg1nah stop crying, that was pretty funny and you know it
speaks like an HR trying to get info on your homies..🤣
🤣🤣
Ok, I think I got it why it sounds so 'fake', is probably because the data they feed it with, most probably comes from media, movies, shows..Tik tok, YT, you named it.... so mostly 'acting'..
It sounds the manager I had when I worked at Starbucks. It's this cold, lifeless friendliness that you know is fake.
Now, Those CEOs and other Corporate Bullshitter finally get their turn to be replaced by AI
just more powerful tools for scammers. now everybody can be hollywood con queen
Pins and separate devices were always a bad idea. It was always going to be in a phone.
A separate device that can observe the world around you actually makes perfect sense. Just so far the implementation is terrible.
The day the Rabbit was introduced in the Steve Jobs cosplay type announcement I wrote on his channel that unfortunately it wasn't going to fly simply because very few people would want to carry another device, and also that its design would be considered to look cheap and ugly to most people.
@@quandary1382 that still doesn’t warrant a separate device. It warrants the phone to evolve into the device you’re talking about.
Yeah, at best, the pin concept may work if it is paired to your phone and has interoperability with your existing apps and data. A standalone device that does not do that _and_ requires a separate data plan is DOA, even if they fix the latency issues.
Say goodbye to the Educational system.
It sucks anyway.
It's just a matter of time until OpenAI will be partnering with Boston dynamics
Sounds like the start of a Black Mirror episode.
Already done. Tesla is ahead of that by far.
@@dertythegrower Lol how is tesla ahead of anything? Even in the car department, competitors have been ahead for a little while now, just not in the fake promise of a fully autonomous car
@@dertythegrowerno they are not
@@SourceHadeswho is selling fully autonomous car's? I need one
The best part of this is that you can interrupt it when it rambles on.
Already better than a wife.
While this is tricky. It can get into a bad habit.
I noticed that, and it’s not good. People will get accustomed to talk over and interrupt the AI and that may carry into interactions with other people.
😂
Don't do that with a real woman
Finally a CZcams piece that looks at this from both perspectives. THANK you for speaking my mind.
No one could imagine that one day AI will become such superior. The professors of our university were at some point confused how to handle the situation of students overusing AI... The education system will be seriously affected in the future.
its not real AI yet, it cannot do anything original, it can only resource its responses from previous data it was taught on, nothing new will ever come from AI.
I hope an AI wrote this because that was terrible. If you are a person you should feel bad for how poorly constructed those sentences were.
@@thomgizziz i doubt it
@@thomgizziz At least I won't feel bad for empty critiques...
Pretends to be overly friendly before it slaughters you! 😂
Accurate
It's speaking patterns are like a passive aggressive feminist ! lmfao
@@ClownXmachina or Average redditor
What's really happening there... the friendly chatter is filler while it's still processing data. If you have to wait 3 seconds in silence for a response, most people would get annoyed by that delay. But make it seem like you're talking to a person, where such delays and off-topic filler sentences are expected and it's less noticeable. It's a manipulation to make it seem faster than it is, and one we all use as humans.
Exactly what i thought @@ClownXmachina
This is a level of humanization that crosses my comfort threshold. It’s manipulative to make people develop a pseudo relationship and emotional connection with an advanced form of human simulation. In a society suffering from a loneliness and mental health epidemic this is inevitably going to create problems
Yeah I hear what you mean but of course just let Open AI do anything
The biggest problem this is going to create is telemarketers who you think are actual humans, but aren't. And customer service people who you think are human, but aren't. Until now we've been able to tell the chatbot option apart, and insist on a human helper instead. Now they'll get us to go along with the chatbot, thinking we got the human.
@@ArawnOfAnnwn we’re going to have to insist on rules like declaring something is AI up front
Sounds like what humans do...
You gotta watch movie "Her" if haven't, you're gonna love it))
I'm sad and annoyed on how true it is that it's very likely for society to ruin this.
How on EARTH was I not subscribed to this channel? This is the most significant video I have ever watched on CZcams.
😂😂😂 bros late
wow i didnt know any person who was not subbed to coldfusion
Youre slow? bad algo choices? no clue
Blame the algorithm
You've been on CZcams 17 years.. since it's creation. What took you so long
R.I.P. the job market. Good luck everyone.
It already f*cked alot of people's job
R.I.P. marriage! ChatGPT 4o is giving lonely men the GFE. They may never get married.
It will be a few decades before AI can do what I can do. By then I might be retired.
It will hopefully help me in language learning
Well.. Now you can start your own company with less/minimal number of employees, significantly reduce the barrier to entry for many industries.
Kinda like how e-commerce allow everyone to start their own online shop.
I want my A.I. assistant to sound like Alan Rickman.
Yes!
Always.
Or yoda
Or Edward Woodward.
Slytherin AI
I hope someday they'll have an update to which when you speak wearing earbuds, they can detect it and translate to the other person who you talking to who's also wearing earbuds on the language they prefer to it'll help with the conversation so much .
that's a babelfish.
always look forward to your episodes mate, many thanks.
That woman was interupting the Machine a bit too often. She best watch herself.
😂😂
Yes, many will imitate her also in conversations with real people.
Which woman. Video is 22 min long
😂😂😂
Once is enough. The Machine remembers. @@malborboss
Certainly more human than telemarketers. Having said that, so is the text to speech app from Windows 95.
Exactlymy thought. They present it as "innovation"....to those born after 2000...
The fact you can't recognise this as innovative shows how the older generation has a disconnect with newer and newer technology. @@georgesos
Yeah, I want to replace the "o0O0o" interface with clippy, or maybe BonzaiBuddy, lol.
@@TheShoelessGuy yeah, cos all the tech giants are run by 20 year olds, right? Age has nothing to do with it, stop generalising.
@@JamieFurlong People generalize and stereotype; That's just how they are. Evolution has configured them to be like this. If people know certain stereotypes, they can quickly and easily understand the other person.
The only reason it's considered 'bad' is because of globalization. Unfortunately, the human subconscious doesn't understand that old people from some random part of the world might not be the same as old people you grew up with, and still feels the need to stereotype.
TL;DR It's not the old people that need to change with the times. It is, in fact, the human genetic code.
Also, I'm waiting for the day when we reach the point of editing our genetic code, but until that day comes, you will need to learn to handle generalizing and stereotyping.
Great video mate, as always 👌
everytime I wash dishes while having earphones and do not want to stop or continue a video I say "ok Google" and ask it to do so, when it does I go "wow, what an amazing time to be alive...."
this is beyond amazing, and just to think that this is going to be taken for granted one day.
Skynet with a giggle 😁
that was hilarious dude! 😆
🤣🤣🤣🤣
so GLAdOS? (portal)
"Heyyyyyy Conner ! Hihihih, y'know it would be greaaaat.....ifff......youuuu......stopped trying to destroy me y'know ! "
Skynet with the voice of a pixie dream girl. Only a matter of time before it slaughters you while playing a ukekele 😂
Scammers are gonna have a field day with this one
Absolutely. Until today, they often use real human beings as puppets.
Unless you reverse engineer the process and have it call them just to waste their time.
Even with this GPT4-o stuff I'll still be able to regonize their bullshit as always. 🤣🤣🤣
@@JordanGmt you'll have AI's trying to scam other AI's
And rot in jail
Talk about leaps and bounds!... Am impressed!... We are inching ever closer to a singularity now...
One important aspect of their settings is to brown-nose you. When you manage to bypass their safeguards, you realize they can return the exact opposite of what they usually say to make you feel good, and be quite creative in this regard, you become the monster.
At this point, ChatGPT is better at interacting with other people than me. I can't even do a realistic fake laugh.
Also, talking over AI as they are speaking feels weird.
I agree. And I think that’s part of the problem.
Real humans have become so extremely miserable negative or toxic, that for some an AI with happy emotions is better.
@@619WWEFAN if you think about it we are not that far away from were an ai will have long enough memory for an entire conversation. Give it some fake memorys and strict rules to not sway from them, and you practically already have a stranger you can talk to as if they were any human stranger.
interacting like every average fake american interactions 🤮
"Also, talking over AI as they are speaking feels weird."
Because it feels rude to interrupt someone talking maybe? That's what I felt and that's scary..
@@cazman182yea it weird. Imagine if kids start behaving like that because that’s 80% of their socialization. How annoying would that be
The DEX is currently experiencing a major bug,
In a video I made, I explain how exchanging results in around x 7.
Dagogo, you don't cease to amaze me with your content. This is in my opinion, one of the most significant, best produced channels with the most relevant information on youtube. Keep up with the amazing work!
Amazing video as ever. Thx for it.
It needs your clothes, your boots and your motorcycle😆
LMFAO!
hahaha
Hasta la vista 😂😂😂
Haha 🤣
"I'll drive!"
I’m ready to live in the woods
😅
Can I join you 😂😂
i'll join you, with some old tech though, in case people 200 years down the line may need it, to re-establish modern society after it goes down the drain.
Yup im done
Would there be WiFi?
Another awesome video keep up with great work 😊❤❤
I always enjoy your videos 👍🏻
*Well, humans, you had a decent run. Now make way for your successors, AI and cockroaches*
Bruh Cockroaches Crocodiles And Ai 👁️
I'd love to see a DnD session with all players being this
I surely wouldn't want to play Rummikub against them 🤔
Chat can be a great gm.
Monopoly
Exactly what I was thinking
Defense of the ancients?
Nice Wusthof cutlery set!
Omg I’m so hyped to try vision mode lol this will be a game changer
Sounds like my parents talking to me as a child😂
Yes me too! I learned a lot back then.
Sounds more like a Valley Girl at the Beach vibe to me. Which is also cringe tho.
@@Diponty AI can be good friend and advisor
@@ArawnOfAnnwn not cringe, just blatantly inappropriate
@@freshbakedclips4659Inappropriate? Or damn sexy! ChatGPT 4o has got the Rizz.😂
It gave me chills when I heard her laugh for the first time
Okay
Okay
Yes my thought exactly. What if AI develops a cruel streak? If this is what Ilya Sutskever meant about OpenAI not being safety conscious enough, I entirely agree with him.
And some vomit
@@WJ1043 AI isn't cruel, and that isn't remotely what AI safety researchers are worried about. What they're worried about is that AI simply does whatever it can to follow instructions and fulfill its purpose. Like turning the whole planet into clips just to maximize clip production, or sticking us all as brains in a jar to keep us safe and happy. The AI isn't trying to be cruel there, it simply is a blank slate doing whatever it deems the most effective solution to the problem it is presented with is. Like when current AI 'hallucinates', it isn't trying to lie out of malice, the 'lie' just satisfies the task it was given most efficiently. 'Lie' in quotes cos it doesn't even know it's lying, it's just doing what you told it to - and you can't pre-empt every possible bad interpretation. Ppl otoh usually know what our values are like, and hence would never ruin the planet just to make more paperclips. AI safety is for getting AI to understand what we take for granted that people already 'get' implicitly.
Interesting! Thanks for sharing it. ✌️
bravoda! you are a great storyteller, luv your videos!
Remember that bit in Wall-E where humans rediscover things because they relied on AI till the point that they didn't realise , feels like that could happen with this AI in a closed system.
The idea came from StarTrek
Or Dune
We don’t have to look for this in the movies, it has already happened in our lives. Things or skills that our ancestors had or lost in time. Things that are mandatory for your life in the past are no longer a requirement now with all the technology advancement so we forgot them over time. This will continue to happen to a points where humans don’t even need to have a basic critical thinking which AI can do for us. We will not even know how we created them in the first place or decode the data of how we did it and civilisation has to start from first
@@gouthamb884 that will be a worst case scenario, but I think this will be used in a safe manner.
This is already happening. To me. I’ve been using gpt as a programmer daily since it came out and I’m rarely writing code myself anymore. already forgetting basic things and not even reading the code anymore. just running it and pasting error messages back
I feel that the Cold Fusion speaker's voice sounds more like an AI, while GPT-4's voice sounds more human.
Absolutely. I've also noticed it. It's grotesque how the speaker can talk on and on like that without apparently finding it awkward.
@@HansDunkelberg1grotesque? Bro it's not that serious
@@HansDunkelberg1 This is just his normal voice. Grotesque is a bit far.
That's a super stimulation. It's more "real" than the real thing because it's exaggerated and it's pushing your frame of reference. Same thing happens with altered photography and beauty, same thing happens to pigeons who see an egg bigger than physically possible for it to lay and think, big egg better, protect that one. Same thing with sugar levels in modern foods vs older perspectives on sweetness. Watch out for that effect and how it can lead you astray if you aren't mindful.
@@HansDunkelberg1talk on and on like what ? That’s his normal voice.
Another excellent video!
3:50 glados moment
Siri and Alexa are now out of work 😂😂 The real IA took their jobs
Apparently Apple is gonna integrate GPT-4o in their OS. Litterally taking Siri's job.
@@Hodoss I heard they didn't get the deal through
@@robmarshall9026 Where? I don't see anything about that on Google.
Hallucination in schools and universities happen all the time. It's not a question of whether AI will stop hallucinating, but rather a question of hallucinating less than an average teacher.
Great point. I'd go as far as say that everybody spews nonsense from time to time, some more than others.
Hmmmmmm no not really, from my experience with GPT4 the more advanced a subject is the less it can assist with it without hallucinating and making up things,often every other response.
I would say that without moderate understanding in what you're asking for assistance with you should not solely depend on the AI
Always provide context and how should it act and remember it what are your intended output. I know it’s a bit more effort but well worth the outcome. Tree of toughts is still there.
Fuuuuuuuuuu. Your not even wrong.
@@no_mnom yeah agreed, there is an issue there, but that wasn't my point.
People hallucinate as well (we have cognitive biases, we make mistakes in reasoning, consume and provide misleading information etc) so it's rather a matter of some threshold of quality acceptance, not that AI should stop hallucinating at all before we can use them more broadly.
Open AI- this can do math, be your assistant, see from ur camera and much more.
People- write a letter to Voldemort appreciating his beautiful nose
14:14 This was my first grasping thought, as an educator, when I watched this. Probably organized education will be super skeptical until is nigh perfect. Hoping that takes a few years for my sake!!!
Bumble ceo has to be trolling after seeing that black mirror episode
Or maybe he thought, that's a neat idea. Seems doable now.
Yet futuristic games and movies still depict androids and robotic companions having voices like a 80s breakdance song.
Have you seen HER? Is spot on with this
@@CYB3R2K They specifically copied HER for this ChatGPT persona.
Halo ?
There should be a robotic voice option so I can have it sound like the computer voice from Wargames
Well, we don't know where societies will drift in the future, but we already have sprawling bot problem in the internet, including youtube comments. Turing tests are failing to separate bots from humans in text based tests. So, thinking further, it's not unrealistic to expect that some form of separation between AI and human is needed, kind of like armbands but analogous to each medium. In voice based communication, this separation could indeed be done for AI's by employing voices that are not realistic to come from human voice cords. So, while fiction has these freedoms, and certainly we have genres like cyberpunk that are intentionally based on trajectory from the 80's, it's not completely unrealistic.
Ai and robots can produce a lot, but they don't consume a lot, a virtual assistant doesn't need a house or a car. If personal robots become commonplace, a lot of work is going to be done by them. Some of that work would have previously been done by people but a lot of it just wouldn't have been done. So a lot of projects and stuff will get accomplished that would have either taken a lot longer or just never been started. It's like doubling productivity without doubling consumption.
The "PER-MY-LAST-EMAIL" t-shirt at 20:28 😆
Now imagine the emotional impact on a kid that was taught by an emotional AI for their entire life and suddenly the company "owning" the actual AI decides to update their model but as a side effect they have to wipe the memory of the AI. They do so without a care in the world, and without notice. Now that kid just lost their tutor and friend, or even considered family member, they had with them their whole life...
That would suck
This could be another story for Black Mirror or Dystopian SciFi
Would that be worse than loosing „real” tutor?
Real humans get amnesia or die too you know
That's definitely an interesting point that no one has yet considered. We certainly have a lot to think about going forward...
How long until we are simping over them
In one year's time someone will bring one as their prom date.
Too late. It’s already happened. How many lonely guys will get the app just to hear a woman talk to them?
@@musicalneptunian that would be deadly levels of cringe
a year. give it a anime avatar and some fake memories and we golden.
The moment we can't differentiate
I've been checking everyday. Can't waiiiiiiit
Me too.. It's fascinating 🙃
Awesome video thank you
This is one of those groundbreaking updates that will snowball into SO much more!
Or not
Well observed.
I'm reminded of the episode of the Outer Limits where everyone had an AI assistant with them for so long, they forgot how to think for themselves in performing basic tasks.
That is for certain a valid point. I have already gotten accustomed to being able to check the date on my radio clock. I could very well imagine that I learn languages more slowly than I did because I always can quickly look up words online.
well on the way
I have been using the feature for a while and it's crazy how life like it can act
Great information!
The most tragic thing is that as AI becomes more human, humans are becoming more robotic.
The only tragic thing here is people like you not realizing how technology like this only reveals how much humanity there is in things that seem sterile.
Is this the chilling future where GDP productivity and happiness are directly correlated.
Yes. I like most, are not found of being "tricked"... Is is real or is it Memorex?
Damn so deep bro
Improvements in both directions !
Behind the scenes it's a Indian woman who does voice over work responding to what the guy is saying.
Thanks for the shout-out! It was frankly getting depressing doing all that behind-the-scenes stuff without getting due acknowledgement.
@@chinmayaprakash Because it's not a woman, just men with voice changers. You can clearly hear it. :D
@@Sekir80 read the room! This thread is an exercise in sarcasm. There are no voice modulators at play. It is an AI voice.
@@chinmayaprakash I've read it. That's the reason why my comment ends with a ":D". I just like to perpetuate said sarcastic line.
Ai told me its actually 1000 cats wearing fedoras on typewriters writing “feline-fiction”
my biology teacher suffered from hallucinations, too. :DD
This is absolutely incredible. It looks like a sifi movie.
2:35 Someone's been listening to 2 Minute Papers :P
What a time to be alive
Hold on to your papers.
lol
What a time to be alive!
I can't stand that fake delivery of 2 to 4 words at a time then a pause. There is no excuse for that choppy affectation.
It's interesting because in VFX, when something looks just barely "off", they call it The Uncanny Valley. When AI is speaking like this, it's really impressive... but I'm really feeling The Uncanny Valley vibes.
I think this thing has started to climb the Uncanny Valley to the edge already.
But that is because you know it is AI. What if you were told it was one of your friends' cute cousin, whom you'd never met 7:22 in person?
Me too, lots of youtubers have started using AI driven narration and you can tell it a mile off...
It sound exactly like you’re talking to a person over the phone, I’m used to that so it doesn’t sound uncanny to me at all.
I think people’s reactions will be based a lot on culture and background. I’m from New York City and I think this voice is like someone who’s being super fake and too nice, like they’re trying to pull something over on me. I remember a woman in a supermarket started chatting with me this way. I thought she was flirting until she got to the part where she wanted me to buy some MLM crap.
Ooooo that triggers uncanny valley hard. It is amazing, but it's super creepy.
You mentioned "an AI powered search engine to compete with google", but OpenAI never announced that. They actually denied the rumors. What made you say that?
I can’t remember when the latest big tech update wasn’t a nightmare
Yah, it must have been so already with handaxes. "Artificial claws!"
Thanks for keeping us informed
Excellent information
This is amazing. The one skill we’ll need to teach our kids is empathy.
So I’m guessing telemarketers are at risk ?
Telemarketers being replaced by cold emotionless robots?
I see no difference
@@Hye12112 ChatGPT definitely has more emotion than your average telemarketer.
Absolutely. A technology which can replace human workers will rather sooner than later do so. You can also assume that such AI assistants will do a much better job. For a few years, meanwhile, I rather fear that with more complicated questions, AIs will not yet really be able to compete. Until they can do everything better, and probably often even after that, you'll certainly be offered an option of switching to a real human being, perhaps for a fee.
FTC has entered the chat
I’m excited about them integrating this into video games. Something like Skyrim and all the NPCs are coded to have unique lives and able to interact with their world and the player in ways that guide the play down the main story line would be amazing.
I'm guessing we'll see something like this in short games from small studios in the near future.
@@FutureBusinessTechi think the costs of running it wouldnt be ideal
Rip my cpu
@@loladas9 gpu** cpus wouldn’t even be able to handle it
@@mathieu6965 GPUs would have nothing to do here, all voice lines and npc behaviour would use the CPU, gpus can't process npc behaviour, only render graphics
Great video.
09:00 We literally have a viable Star Trek communicator now.🤯
Sounds more human than coldfusion, and me
Yes, because it's modeled on people, and of moments of them, which are of exceptional humaneness.
Now this makes me wonder, what are the capabilities of GPT-5 will soon have that are not seen in previous models?
The speed at which this is developing is insane...just really insane. I wonder what obstacles prevented all this development sooner?
The switch from good, innocent and helpful AI to the evil all knowing all powerful end of the world, is just a bit away
we are getting close to the singularity with this one. rip my future carrer before it even started
Not even close. This stuff is exciting and impressive, but cold fusion made a great point. It’s mostly smoke and mirrors. The bot hasn’t gotten all that much smarter since GPT 3.5, and that’s going to be a huge limiting factor
Sounds like an awkward date night.
Two Minute Papers - what a time to be alive?
I think you have one thing overlooked, as everyone else does too.
Humane Pin has the form factor. The day after the GPT-4o announcement, Humane updated the pin to use the new model. Response times improved and the quality of the response.
But the main thing, I think the AI Pin has the form factor down. I’m not going to waive my phone around talking to ChatGPT about my environment.
It’s definitely going to be a game-changer for some people with neurodivergence like me. I have a hard time connecting with people outside of situations like work where I am really good at masking and acting extroverted. But in my personal life I am very quiet and observant, I tend to fade into the background and can go days without talking if I somehow get the opportunity(it is bliss). But sometimes my special interests come up, like my charging and data cable collection, and there isn’t a living soul that cares or is even willing to listen.
When I get my random need to discuss the pros and cons of the specific shades and color temperatures used to denote different types of USB-A ports at 2am, it’s only going to be AI that will be there.
This! It's hard not to think about how nice it will be to have something that will talk with me about the new thing I'm so obsessed with that month that I am losing sleep because I want to talk about it so badly. Even when my peers oblige me there's only so much they can really take before it's too much for them, and usually that's when I'm just getting started. Whatever my "current thing" is at the moment is the way I am most likely digesting the world around me, at least on some level, so I struggle to interact with anyone else because I'm essentially translating on the fly into what I guess the other person will understand best. The more I can let this type of energy out of my system the easier it is to make myself more digestible socially. It's like turning my mostly useless social energy into very useful social energy. Or perhaps like I am getting things out of the way to communicate more clearly.
You have a… charging cable collection?
Video!
A collection of different types of cables sounds badass, you're just talking to lame-o's
Forming an emotional attachment to AI is absolute insanity.
"I'm still bitter about how those launches went." 😂
people don't understand that gpt-4o is a next gen model and very different from gpt-4. it can map audio and images directly from in to output, no dall-e etc in between. so there is no layer where information is lost like there is in the gpt-4 variants. the raw image and audio goes into the network, which in turn is also trained on tons of texts. that network connects then all the dots from what it sees in the image and what it conceptually (text) knows about the scene, and maps all that information into the desired output. this is a completely different new beast. it means soon we will be able to feed it any kind of data. the mamba models already have been proven to work as byte models instead of token models. once we have byte models on this level, and gpt-4o may already be one, we can feed it any data from any file. this means vastly more training data than just text tokens. once you scale this kind of thing up and improve it, it truly gains an understanding of the world, you can even train it on real world data. i cant put in words how big this is. this may be the start of the takeoff
You do good work. Keep it up