What’s really amazing is people don’t even look at any of their photos going past maybe a day or two. Are used to be people would have get-togethers and look through slides of their vacation or way back in the day today. It just falls away and people forget about it.
The pie shop demo had some areas of concert for me. The prompt specified the budget was under £1000 yet the AI accept the cost of the pie at exactly £1000. When the shop keeper said the pies would be 37secs to deliver that should have raised a red flag and there should have been follow up questions to verify authenticity of the order.
I, as a European, have had access to O1-Preview at the same time as the Americans, so I have no idea what they're talking about. edit: it appears this was about API access. edit2: one UK commenter claims to have had API access for a while, so the original confusion stands.
Just want to point out that everything in these demos is available right now, no waiting period for any of it. o1-mini and o1-preview are available right now for Tier-4 and Tier-5 API users and the Real-Time API is available for all tiers. (could be some local restrictions, but everything is available)
This precedent of letting a powerful AI model livecode a program to control an actual untethered machine in front of a crowd of people and then execute it without any testing has a lot of potential.
I'm confused, was Wanderlust developed by o1 or is that o1 controlling a 3rd party site with agenetic abilities? The planet demo is a 3rd party app utilising o1's API? Or it's an app generated in the GPT chat window ?
This "show me some restaurants and place an order" demo is so passé. What is painfully obvious (and I am dating myself) is "Europe on $5 a day", a book I carried around while traveling over 50 years ago provided a better user interface, more privacy, a quicker response and it almost always worked (sometimes it was out of date, but LLMs can have the same problem). Nobody has (yet) done a demo of an agent I would actually use.
A childish response versus trying to understand what he’s stating… He hit his spot on with the demo being passé… And for all this over hype of agents, no one really has shown anything groundbreaking, other than demos
Ok, but what they have achieved is things that, 5 years ago, I thought were at least 25 years in the future. (as did most people who followed such things).
🇧🇷🇧🇷🇧🇷🇧🇷👏🏻, I'm eagerly awaiting the release of real-time video for Plus users from OpenAI as well, as it was originally mentioned as part of the ChatGPT Omni update, which sadly never reached us. This feature will be revolutionary, enabling us to tackle a wide range of daily tasks more efficiently. Real-time video integration within ChatGPT would greatly enhance productivity by allowing for interactive, dynamic assistance and more streamlined workflows. It would be especially useful for tasks like desktop sharing-being able to visually assist and collaborate on real-time activities is just phenomenal. I hope this feature rolls out soon, as it could drastically improve how we approach everyday challenges.
Hi, the demo of the drone was done with o1-mini, if I recall correctly (using cursor as well). No o1 was demonstrated in London-only the o1-mini and o1-preview.
@@nikhil_jadhav or… or… not be wage slaves who have to focus on repetitive labor simply to stay alive. I think there are a lot of people who miss how much human capability and creativity we leave on the table based on the way that we structure our current society.
im sure the best thing that we've gotten from openAI doesnt perform as well as the worst thing they've actually shown off in these demos. It doesnt matter what they are developing we will not get any of it.
@@Saif-G1 but with the real time api Is so fast I think they found the perfect combination This idea they made is not really original I see many guys in TH-cam create really powerful agents But there design is so good and the phone ai calling give it a very futuristic vibes
@@retrobossarcade3524 I'm from Germany, and we had access to the o1 preview model when it was officially announced on day one, probably around 3 hours after the official announcement. So to answer your question, we've got it for a while.
@@andreaskrbyravn855 Let's hope the release the full model. Should be pretty compute heavy and expensive to run from what i've read online. Meaning, ChatGPT Plus users most likely get a pretty small message cap.
Europeans have had access to O1-Preview. We have been receiving some of the advanced voice interactions lately. The preview is for paying customers only by the way and there is a limited number of requests you can do.
WHAT!? Are you saying those graphs and visuals are generated in real time as you speak with the AI? This is absolutely insane... This is revolutionary for learning! I am studying sociology and I am really looking forward to trying this.
They are going to limit and decide what is research. If it does not fit accepted theory and open ai internal goals it will not help or not help to its full capacity. I’ve been working on complex things and overall preview is far worse for this than mini preview , it’s being censored more. I think I’m lucky I did the bulk of my work already before this really started. Cause others will be severely limited actively by the very tools they need to break new ground.
@@travisgoesthere prove true ? It’s just what it is bro. O1 has that pre reasoning dialogue. It checks in there and yes has put out responses that the work is not in line with accepted theory. I have seen it say pseudoscience in the pre dialogue, and checking code for anything agains open ai policy. Things like that. Worse in o1 than o1 mini. And a trend of not getting deep into the work as o1 mini or 4 or 4o. More rejection of requests for complexity etc.
@@travisgoesthere saying misinformation like it’s supposed to taint facts is a bad look. 👀 who u work for? Or ya just that stupid no work bias needed ?
That would be super-useful to me, as I'm an expat who hasn't yet mastered speaking the native language (Portuguese). I'd love to have an assistant speaking on my behalf to business people in their native language.
Gen-X. The generation without a face. - They're taking the "order lunch for the audience trick" on tour! - the realtime API is like having an expert tutor in the room with you 24/7. "Game changer," feels like an understatement.
In January I preorder Rabbit R1 for it's advertised LAM capabilities. I received it in early june. It already sits on my memorabilia shelf with other stuff from the past. The new releases in AI made that device obsolete before really getting adopted. Now we can just build our complete own apps.
Dude, just focus on programming, the improvement is on programming and math. People dont need Chat GPT to locate their favorite restaurant, they use Google Map for that
I live in Poland, and I had access to the 01 preview in ChatGPT and as the API at the same time as people in the U.S. :) Maybe they're talking about access to the 01 model.
Try to fly a drone without having a premade software and controller for it. You will have to understand, what kind of signals you need to send for controlling the drone, code a Controller-Interface with Buttons for Movement, setup a live Video Connection, and on the way probably fix 100 errors. Without AI this would take someone without experience, hours to days. Being able to do this without major errors in a couple of minutes is absolutely insane.🤯
@kirschdieblp4242 ohh ok, so all the balancing and stuff to make it say still was made by the ai? Yea thats impressive, I have actually tried that, had a hubsan speeddrone without any software, it was completely impossible for me to learn to fly more than a couple of seconds before i crashed 🤣
That's because you can't define Gen x. Or a bunch of badasses coming from a period in time where we got outside roamed around got into trouble and actually did things.
I expect o1 to become some kind of enforceable legal service standard. Ex1: hospital service, customer service, housing service and gov services in general. If o1 can't get through, it's time to sue the provider for being a total duce lol
Matt that demo is straight from OpenAI RealTime console... just with few extra functions, anybody can get repo and add whatever he wants (to which rest api he has access) that sucker is expensive that demo alone burned around £30 quid trust me...
This was a bit click baity. First of all, it looks like o1-preview, not the full o1, so the title is off. And why did you not analyze why they would use o1-preview for realtime interactions? It's not real time. Just confusing and not a lot of worthwhile content like normal.
So the next time the AI orders 200 real pies and a pie shop tells them it's going to take 4 hours to deliver, the AI Is going to flip out and say that's totally unacceptable since 200 pies can be delivered in 43 seconds....
Dear lord. Planning a trip - FFS, this has been done to death. Cure cancer or, really, pretty much anything else except suggest the Raffles *&%!!ing Hotel when planning a trip to Singapore. What next? Things to do in New York: Have you thought about a pointless ten minutes standing in, wondering why anyone goes to, Times Square? How about the Eiffel Tower in Paris? Even GPT itself has this to say: "It’s endlessly baffling, isn’t it? It’s as if the AI marketing team can only fathom us using this monumental technological breakthrough to build itineraries that would bore even the algorithm that came up with them. Madame Tussauds-for the love of all that’s artificial! If they’re going to sell AI as transformational, surely they could flex its muscles on something more novel than suggesting overdone attractions. It could be that they’re aiming for "relatable" by hitting the generic highlights, but it’s painfully underwhelming. The irony is that AI could actually tailor travel ideas to one’s real tastes (think obscure 3 a.m. jazz clubs in Berlin or Scottish ruined castles with a haunted reputation), yet they go the route of an uninspired guidebook. For once, it'd be great if an AI demo threw something unexpected into the mix, perhaps sparking an adventure you'd actually want to write home about. But alas, “Hey, did you know the Tower of London exists?” "
Hey AI boy, there are tools to clean up audio you know. Why am I listening to this echo chamber sound from the presentation? Try and find a sponsor to fix your audio maybe
Well, I'm not very impressed with ordering pies, but I guess the real thing here is the planning and then execution with a voice interface. That went well, up until the pie guy went off script and said they'd be there in 43 seconds. It sounds to me that this is accepted instead of questioned. But more to the point getting a delivery time for 1000 pies should have been part of the planning process, or better it should have been questioned before placing the order. A human would have realized 1000 pies will take time to prepare, that they are not meant for now, and would have asked about timing before placing the order. Feels like a mistake along the lines of the old problem of shirts drying on a line mistake from before. I wonder what would have happened if the pie shop answer was more realistic and took it as an annoying prank call.
isnt this the exact demo they did when they announced 01-preview?? Like literally word for word shot for shot? This is a recycled demo from a month ago??
I can't wait for a REAL world demo where the person is talking quickly, with an accent, cutting them off, talking quickly, using slang, colloquialisms, and the AI is just horribly lost
Is it just me, or is all the stuff that was presented really not that big of a deal? For every single example you've presented there's almost an open source alternative, and it's not a big deal, i.e., it's not impossible to program them... I wonder... what's so amazing about that?
whats even worse about this situation is the not very impressive stuff they are showing off isnt even the final product they are giving us so its even LESS impressive by the time we get our hands on it.
@@neoglacius Nope, try to assure that any of the frontier models achieves 99% accuracy (which is still low for real-life problem solving because in that one percent that the model fails it may cause you huge financial losses for example or give you an incorrect advice but that's what the people do and brag about on the Internet so the model will repeat it), and will not confuse everything with everything else. If you can't then there has to be someone to correct these mistakes. So It's still a long road ahead of models but sure, it get shorter and shorter every day.
@@szlagtrafi9115 *models achieves 99% accuracy* yeah bud, but if it gets 50% of accuracy this means that 1single dude can 'check' this 99.9% accuracy, so if 1 can do the job of 50 other white collars so all those jobless white collars wont buy in small-middle businesses collapsing the whole economy
Its negligent towards humanity to sell powerful AI chips to India when you know they will resell them to anyone. Russia just bought 1000 super powerful American AI chips from a shady Indian company. Whilst im for AI freedom and the companies that create it, including open source, we need to keep a lid on the hardware. This technology is presenting a level of power never seen before in the hands of anyone who has the hardware.
The thing is, Open AI has never had issues showing us awesome demos. They just never give us exactly what they demoed
In the coming weeks.... /s
@bigglyguy8429 hopefully we get the same version that they demoed
Yeah... when do we in Europe get the "new" voice chat function?
Sounds like most tech companies
@@bigglyguy8429 In the next year, probably... at the end of it
Seeing people take pics of presentations is like watching someone film a concert with a toaster.
Is this an insult to us who can only afford toasters?
Keeping toasters employed in 2024.
AI toasters will replace pics of presentations.
especially when it's out of focus for half the time
What’s really amazing is people don’t even look at any of their photos going past maybe a day or two. Are used to be people would have get-togethers and look through slides of their vacation or way back in the day today. It just falls away and people forget about it.
The pie shop demo had some areas of concert for me. The prompt specified the budget was under £1000 yet the AI accept the cost of the pie at exactly £1000. When the shop keeper said the pies would be 37secs to deliver that should have raised a red flag and there should have been follow up questions to verify authenticity of the order.
Its just a tech demo, devs are responsible to implement adequate guardrails
I, as a European, have had access to O1-Preview at the same time as the Americans, so I have no idea what they're talking about.
edit: it appears this was about API access.
edit2: one UK commenter claims to have had API access for a while, so the original confusion stands.
Same, but its a restricetd ammount of prompts for week. maybe a change to that...
Same here. Maybe this is o1-preview + computer use
How do you mean?
Yes, and this is a demo of o1, not o1-preview, they are not the same
@@Tr1gg3e he said o1-preview end of video
Just want to point out that everything in these demos is available right now, no waiting period for any of it. o1-mini and o1-preview are available right now for Tier-4 and Tier-5 API users and the Real-Time API is available for all tiers. (could be some local restrictions, but everything is available)
im from Belgium, Europe and we had o1 prev same time as you. i absolutely love using it
This precedent of letting a powerful AI model livecode a program to control an actual untethered machine in front of a crowd of people and then execute it without any testing has a lot of potential.
lol
since the drone flew, lets call it...skynet
This is like watching a rapper do a “freestyle” which is very obviously prewritten
I'm confused, was Wanderlust developed by o1 or is that o1 controlling a 3rd party site with agenetic abilities? The planet demo is a 3rd party app utilising o1's API? Or it's an app generated in the GPT chat window ?
This "show me some restaurants and place an order" demo is so passé. What is painfully obvious (and I am dating myself) is "Europe on $5 a day", a book I carried around while traveling over 50 years ago provided a better user interface, more privacy, a quicker response and it almost always worked (sometimes it was out of date, but LLMs can have the same problem). Nobody has (yet) done a demo of an agent I would actually use.
And nobody cares what you would use , grandpa
@@travisgoesthere And nobody cares about your proof of an ahole. 💩
A childish response versus trying to understand what he’s stating… He hit his spot on with the demo being passé… And for all this over hype of agents, no one really has shown anything groundbreaking, other than demos
Another Boomer screaming at technology 🤷♂
@@Mrbriangalvan 😅🤣😂 off to bed grandpa
OpenAI demos what they want to achieve, not what they have achieved
That was all running on API's and code available today.
Ok, but what they have achieved is things that, 5 years ago, I thought were at least 25 years in the future. (as did most people who followed such things).
what they have achieved and what you never will
EU has o1-preview.
🇧🇷🇧🇷🇧🇷🇧🇷👏🏻, I'm eagerly awaiting the release of real-time video for Plus users from OpenAI as well, as it was originally mentioned as part of the ChatGPT Omni update, which sadly never reached us. This feature will be revolutionary, enabling us to tackle a wide range of daily tasks more efficiently. Real-time video integration within ChatGPT would greatly enhance productivity by allowing for interactive, dynamic assistance and more streamlined workflows. It would be especially useful for tasks like desktop sharing-being able to visually assist and collaborate on real-time activities is just phenomenal. I hope this feature rolls out soon, as it could drastically improve how we approach everyday challenges.
Hi, the demo of the drone was done with o1-mini, if I recall correctly (using cursor as well). No o1 was demonstrated in London-only the o1-mini and o1-preview.
humans, we had a good run!
nahhh
Glad it’s over honestly. Been exhausting.
Finally we can relax and hopefully give up achieving, learning, researching and then rest in peace. Let the AI take it forward hence forth.
@@nikhil_jadhav or… or… not be wage slaves who have to focus on repetitive labor simply to stay alive.
I think there are a lot of people who miss how much human capability and creativity we leave on the table based on the way that we structure our current society.
Happy to pass the torch
This is unwatchable. Why isn't there an official demo video of this? Do they assume that all devs were at this talk in London?
im sure the best thing that we've gotten from openAI doesnt perform as well as the worst thing they've actually shown off in these demos. It doesnt matter what they are developing we will not get any of it.
Its all D a R P A tech on a role out anyway.
Can someone explain what happened in the video, i didn't understood anything was that an agent or clone or what please.
first of all, learn english, second of all, o1 is an ai that can think before giving responses
Agent powered by O1 preview and real time api
Really! o1- preview is smart but slow
@@Saif-G1 but with the real time api
Is so fast
I think they found the perfect combination
This idea they made is not really original I see many guys in TH-cam create really powerful agents
But there design is so good and the phone ai calling give it a very futuristic vibes
I hope it will be available in November.
And hopefully on android and iOS not just for bc
In live in France , i had access to 01-preview day 01
Live now or for a while?
@@retrobossarcade3524 I'm from Germany, and we had access to the o1 preview model when it was officially announced on day one, probably around 3 hours after the official announcement. So to answer your question, we've got it for a while.
its not preview they talk about its full 01- its better at everything.
@@andreaskrbyravn855 Let's hope the release the full model. Should be pretty compute heavy and expensive to run from what i've read online. Meaning, ChatGPT Plus users most likely get a pretty small message cap.
Europeans have had access to O1-Preview. We have been receiving some of the advanced voice interactions lately. The preview is for paying customers only by the way and there is a limited number of requests you can do.
We had access to o1 preview I think the same time as you guys. Remember still the live stream with you and I think I just also had it or a day later
WHAT!? Are you saying those graphs and visuals are generated in real time as you speak with the AI? This is absolutely insane... This is revolutionary for learning! I am studying sociology and I am really looking forward to trying this.
What you don’t understand is there is no point of studying.
@@kidooaddisu2084 Why not?
@@kidooaddisu2084 because AGI 2026?
but even if robots & AI do all our work after 2030, one still can study as a hobby
@ is it smart wasting all your time learning long division when calculator is already invented?
@@Blue_nip He's just joking. That or he isnt worth listening to.
Now that algorithms can build software, we as humans must be resilient and think of new jobs for us to have.
I'm using Mammouth in London and I asked o1 one question, which was incorrectly answered. Claude Sonnet 3.5 is the best so far
Up until 2 days ago, o1 allowed you to analyze and improve on "machine learning model code." Now... it says you're violating OpenAI policies. 👀
They are going to limit and decide what is research. If it does not fit accepted theory and open ai internal goals it will not help or not help to its full capacity.
I’ve been working on complex things and overall preview is far worse for this than mini preview , it’s being censored more.
I think I’m lucky I did the bulk of my work already before this really started. Cause others will be severely limited actively by the very tools they need to break new ground.
@@M-dv1yj Any way you can demonstrate tht to be true or just spreading misinformation?
@@travisgoesthere prove true ? It’s just what it is bro. O1 has that pre reasoning dialogue. It checks in there and yes has put out responses that the work is not in line with accepted theory. I have seen it say pseudoscience in the pre dialogue, and checking code for anything agains open ai policy. Things like that. Worse in o1 than o1 mini. And a trend of not getting deep into the work as o1 mini or 4 or 4o. More rejection of requests for complexity etc.
@@travisgoesthere saying misinformation like it’s supposed to taint facts is a bad look. 👀 who u work for? Or ya just that stupid no work bias needed ?
@M-dv1yj address the original statement, Gomer
I think it's a demo of how should be the interaction with every kind of application through LLMs when they will integrate them
That would be super-useful to me, as I'm an expat who hasn't yet mastered speaking the native language (Portuguese). I'd love to have an assistant speaking on my behalf to business people in their native language.
No, EU European have already access to o1-preview since months like in the US. I guess that is also true for the exile Europeans from GB. ;-)
Will shops accept orders from bots? Will the contracts be legally binding?
I'm having a freken dejavu, especially with drone navigation and booking something. Wasn't it the same demo cases when they introduced o1-preview?
Wondering if being I’m an upgraded user now if o1 will be another upgrade fee or we will get o1 at the current price?
It was an awesome DevDay!
I'm in Europe and we had access to o1 preview too.
I didn't understand how o1 is integrated into Wanderlust?
Or is it not? And just viewing the screen??
Gen-X. The generation without a face. - They're taking the "order lunch for the audience trick" on tour! - the realtime API is like having an expert tutor in the room with you 24/7. "Game changer," feels like an understatement.
In January I preorder Rabbit R1 for it's advertised LAM capabilities. I received it in early june. It already sits on my memorabilia shelf with other stuff from the past. The new releases in AI made that device obsolete before really getting adopted. Now we can just build our complete own apps.
Yeah we've had access to o1-preview in the UK for a while now. Must be just that they got feee access for attending.
When will us common folk get API access to o1-preview ??
Everyone enjoys the rapid development of artificial intelligence, so no one has the right to complain about the dangers.
I can see people at Open AI using Arc Browser and Cursor 🔥
8:59 sounds out of a movie 🤯
All of these demos seem contrived, much like their voice demos some months back that they still haven't fully delivered on.
they have delivered on it. I use it every day
You can’t spell slaughterbot without laughter
I have access to o1 preview, but it can't locate even a restaurant nearby. It asks me to search on a map software. What's happening ??? 😮😮😮
Dude, just focus on programming, the improvement is on programming and math. People dont need Chat GPT to locate their favorite restaurant, they use Google Map for that
We europeans had access to o1-preview already. They were talking about API access.
Maybe they meant o1-preview is soon to be available to free tier users? It currently isn't, so...
Not preview maybe mini
I live in Poland, and I had access to the 01 preview in ChatGPT and as the API at the same time as people in the U.S. :) Maybe they're talking about access to the 01 model.
Learned some space facts. 10/10.
I’m in Spain and I’ve been using o1 preview for a month or so…
It's becoming more and more human-like every day.
1:40 Looks like GenZ is "Back to 60s/80s"
GenX is definitely the "Grunge shirt" generation, with a streak of Ted Kachinsky lol
I'm in Ireland and have o1 preview.
It would never manage to book a ferry in Greece, for sure.
Where is the info?
I have a same feeling as after the Tesla day....Yes, it is probably the feature, but not tomorrow, for sure.
Create a drone interface feed the interface to the Ai. Ask the Ai to spew the program I previously fed him.
I did not get the drone demo? Whats the deal? It flew up did a spin and down again.
Try to fly a drone without having a premade software and controller for it. You will have to understand, what kind of signals you need to send for controlling the drone, code a Controller-Interface with Buttons for Movement, setup a live Video Connection, and on the way probably fix 100 errors. Without AI this would take someone without experience, hours to days. Being able to do this without major errors in a couple of minutes is absolutely insane.🤯
@kirschdieblp4242 ohh ok, so all the balancing and stuff to make it say still was made by the ai? Yea thats impressive, I have actually tried that, had a hubsan speeddrone without any software, it was completely impossible for me to learn to fly more than a couple of seconds before i crashed 🤣
Wait, for the planet demo. The Ai generated the planets at the same time? 🤯
We did have o1-preview/mini here in Europe.
I remember google/microsoft showed something like this a while ago? Pie ordering
Pie ordering. It's really pitiful. Probably the US military has a version that can decrypt 2048 bit security. Meanwhile, the public can order pies.
Looks amazing, but still extremely expensive for common, everyday use as an assistant.
Europeans also have access to the O1 preview at the moment.
That's because you can't define Gen x. Or a bunch of badasses coming from a period in time where we got outside roamed around got into trouble and actually did things.
I expect o1 to become some kind of enforceable legal service standard.
Ex1: hospital service, customer service, housing service and gov services in general.
If o1 can't get through, it's time to sue the provider for being a total duce lol
RealTime API has too low volume to use it in production :(
when is o 1 coming
emm... google do that in 2018... "Google's AI Assistant Can Now Make Real Phone Calls"
I wonder what will become of OpenAI once X has 5 times the Compute.
I still don't have API access with a tier 3 level. Very annoyed. From South Africa so probably just hasn't been opened up to us yet.
Matt that demo is straight from OpenAI RealTime console... just with few extra functions, anybody can get repo and add whatever he wants (to which rest api he has access) that sucker is expensive that demo alone burned around £30 quid trust me...
This was a bit click baity. First of all, it looks like o1-preview, not the full o1, so the title is off. And why did you not analyze why they would use o1-preview for realtime interactions? It's not real time. Just confusing and not a lot of worthwhile content like normal.
So the next time the AI orders 200 real pies and a pie shop tells them it's going to take 4 hours to deliver, the AI Is going to flip out and say that's totally unacceptable since 200 pies can be delivered in 43 seconds....
Didn’t Google do the same thing back in 2018? I don’t know what the big deal is.
Those in control decide who gets popularity (even if the tech is old).
we do have a preview in the UK I have had it for a while now
Open AI using Arc browser Nice
Dear lord. Planning a trip - FFS, this has been done to death. Cure cancer or, really, pretty much anything else except suggest the Raffles *&%!!ing Hotel when planning a trip to Singapore. What next? Things to do in New York: Have you thought about a pointless ten minutes standing in, wondering why anyone goes to, Times Square? How about the Eiffel Tower in Paris? Even GPT itself has this to say:
"It’s endlessly baffling, isn’t it? It’s as if the AI marketing team can only fathom us using this monumental technological breakthrough to build itineraries that would bore even the algorithm that came up with them. Madame Tussauds-for the love of all that’s artificial! If they’re going to sell AI as transformational, surely they could flex its muscles on something more novel than suggesting overdone attractions.
It could be that they’re aiming for "relatable" by hitting the generic highlights, but it’s painfully underwhelming. The irony is that AI could actually tailor travel ideas to one’s real tastes (think obscure 3 a.m. jazz clubs in Berlin or Scottish ruined castles with a haunted reputation), yet they go the route of an uninspired guidebook. For once, it'd be great if an AI demo threw something unexpected into the mix, perhaps sparking an adventure you'd actually want to write home about. But alas, “Hey, did you know the Tower of London exists?” "
Wow. Really puzzled how they are able to make speech latency AND interaction work with such low latency. Even for Open AI that is really impressive.
It's combined into 1 model
Hey AI boy, there are tools to clean up audio you know. Why am I listening to this echo chamber sound from the presentation? Try and find a sponsor to fix your audio maybe
Well, I'm not very impressed with ordering pies, but I guess the real thing here is the planning and then execution with a voice interface. That went well, up until the pie guy went off script and said they'd be there in 43 seconds. It sounds to me that this is accepted instead of questioned. But more to the point getting a delivery time for 1000 pies should have been part of the planning process, or better it should have been questioned before placing the order. A human would have realized 1000 pies will take time to prepare, that they are not meant for now, and would have asked about timing before placing the order. Feels like a mistake along the lines of the old problem of shirts drying on a line mistake from before. I wonder what would have happened if the pie shop answer was more realistic and took it as an annoying prank call.
Uh. This may be pie in the sky for most of you but rich people are building this for themselves. You will work within it.
We will be so manipulated on infinite levels
Europeans already have access to o1-preview
We've had o1-preview in ChatGPT in the UK since it was launched so I'm not are what that is about.
isnt this the exact demo they did when they announced 01-preview?? Like literally word for word shot for shot? This is a recycled demo from a month ago??
My God, it reminds me almost like the beginning of slaughter bots!
The o1 and realtime voice apis are very costly right now
I can't wait for a REAL world demo where the person is talking quickly, with an accent, cutting them off, talking quickly, using slang, colloquialisms, and the AI is just horribly lost
the first video of the realtime api is old af. that bakery man ordering was featured way too long ago.
Nice
Is it just me, or is all the stuff that was presented really not that big of a deal?
For every single example you've presented there's almost an open source alternative, and it's not a big deal, i.e., it's not impossible to program them... I wonder... what's so amazing about that?
whats even worse about this situation is the not very impressive stuff they are showing off isnt even the final product they are giving us so its even LESS impressive by the time we get our hands on it.
@@GravLabs-to6xy U.S. elections. The losing party will be too hasty in shifting the blame on new exponential emerging technology.
Say hello to on-demand software!
Ps. UK had access to the o1 preview same day as the US
nope, say hello to on-demand any white collar job, try plumbing while it last
@@neoglacius Nope, try to assure that any of the frontier models achieves 99% accuracy (which is still low for real-life problem solving because in that one percent that the model fails it may cause you huge financial losses for example or give you an incorrect advice but that's what the people do and brag about on the Internet so the model will repeat it), and will not confuse everything with everything else. If you can't then there has to be someone to correct these mistakes. So It's still a long road ahead of models but sure, it get shorter and shorter every day.
@@szlagtrafi9115 *models achieves 99% accuracy*
yeah bud, but if it gets 50% of accuracy this means that 1single dude can 'check' this 99.9% accuracy, so if 1 can do the job of 50 other white collars so all those jobless white collars wont buy in small-middle businesses collapsing the whole economy
AI will either make us super rich or destroy the world.
Why cant both be true simultaneously?
No, I'll see YOU in the next one.
I think it will come here in november
First time I've seen an AI haggle for a price. lol XD
Its negligent towards humanity to sell powerful AI chips to India when you know they will resell them to anyone.
Russia just bought 1000 super powerful American AI chips from a shady Indian company.
Whilst im for AI freedom and the companies that create it, including open source, we need to keep a lid on the hardware.
This technology is presenting a level of power never seen before in the hands of anyone who has the hardware.
hype
the amazing part of this demo is that o1 could understand his Dutch accent. :)
as usual, these will come out to paid users in about 9 months ... (after other companies will release their versions)
In Europe o1 preview is longer available