Just some kind of space case for blind people, because it helps them cook, helps them, uh, clean.
Everyone understands that NVIDIA has the coolest chips, but, uh, Grok received some incredible performance inference LLaMA-a-a-third of 8 billion parameters.
Universal income will not be in the money, but in the fact that a person will be given the oppressive number of power to use artificial intelligence.
Hello, everyone! We're on ToTheMoon. We're going out once a week with news from the Silicon Valley, with technological news around the world. We're gonna get different ins. Always four. Hey, guys.
Hi
.
This week, much of the update took place in the OpenAI and ChatGPT and GPT in general. Uh, I want to start, uh, about, uh, Open, Financial Times and the book archive that destroyed OpenAI.
We've been talking about this subject for a number of episodes. It is primarily linked to the fact that the developers of the genetic and chat-bot makers use the dates, use the data available on the Internet. And, uh, these data can be protected by copyright, yes. For example, it could be, uh, information in, uh, encyclopedias, it could be information in some newspapers, magazines, some kind of forums, somewhere else. And accordingly, these data are used for model training. After that, the companies that created models, uh, profit from the product they made, but they do not share profits with, say, the data sources they used. And there were a lot of time, uh, litigation on that, yeah. And there was a pretty big news, uh, that the OpenAI had a deal with the Financial Times, where they, for example, formalized their cooperation. The news said that now, uh, ChatGPT, from some point on, it would even leave a link to the Financial Times site, and the Financial Times seems to be just withdrawing some of his claims. Anyway, they agreed. Second news on the same subject was the use of the ChatGPT book archive. And the OpenAI here was so stiff. They said that, uh, this book archive was used a long time ago. Uh, in the current version, uh, uh, in no way does it affect. There, GPT 3, seems to have been trained in this book archive. The staff that did it had been resigned for a long time, the company's not working, and the book archive we've already removed, yes. Nothing is present, like we're getting a smoothie. And it's interesting here, actually. On the one hand, it's kind of like, yeah, so we talked a lot about revenue share. So, uh, Sasha Mashrabuch, you talked about it too much, yes, which is kind of good if companies share each other. And on the other hand, I see it as, uh, a big gap, uh, for all the next companies, the next start-ups. This way will be much more difficult than the OpenAI did in time, yes. Well, in a rude way, just, um, a pirate gathering the dates. Now, just, well, we decided that okay, we agreed, that's not gonna be a thing for us. But other companies that catch them are probably not gonna make it, yes, because the market changes. Here. Is it good that the Financial Times with the OpenAI agreed or did this story unjust from the perspective of other market players, and the market had to work equally for all?
I'm trying to find information about de-de-d-did deals. Oh, and they also agreed with Reddit, yes. I think the main problem is that all these conditions are not transparent. I think Google made more transparent terms of payment to the cryographers and payments, website payments. And here it is all on-- and this is not going transparent, so it's not clear whether they, whether the OpenAI really wants to do it, make some standards, or they do it to get them back. That's the information that's in the public now, I'm very hard to say.
These news seem to have just been the most complex-- the most complex. It's not a mass phenomenon, but it's just, apparently, the most difficult situation there with the Financial Times, say, for example. And that's how they closed the question. But it's not a mass agreement with all the media that's the same or the same data sources on the Internet. That's how I see it. So here I, uh, rather tend to think that I closed a private case, and that just came to light.
Well, just like Google and the Internet, they somehow led to some kind of democratization, yeah, that a more-a less-love person could have tried to be a yeller on YouTube, and whatever big he is, small, is If he's got a reputation or not, if he gets a check on the quality of the view, yes, there's-- and it seems to me that there was a good movement. Google did a lot for a democrat-- well, to democratize the creation of content and so on. And OpenAI is like moving in the opposite direction that they only negotiate with big ones, with those who can make a few million dollars to judge OpenAI.
I think there's still an interesting story, that... Well, first of all, I'm gonna support everything that's said, and I'm gonna say that , uh, is now going on, uh, new, uh, new functionality, new ties, all over the world. And, in fact, every time we get something new, we'll talk about it, uh, that's the opportunity, uh, that's the updates of the GPT-4o, which are the four of which we've come out. Every time something new comes about, it's a huge number of different discussions. It's a little intermediate. And, in fact, this is all-- all of the story that deals with intellectual property, rights, possession of content, all of the references and details, it's now, uh... I understand that no one understands what happens in the future, how it's gonna be regulated, uh, who, where, uh, who's gonna check on how it's gonna be checked. Not enough power to actually check everything. And, uh, even the plastic doesn't check in the systems. Now, um, they're updating the anti-plugia system, they're updating a lot in Instagram, for example, yes, and Facebook, Instagram. It's not even a crypt. What to say about, uh, thing, I mean, a plastic, which is clearly visible, clearly visible. What to say when there's artificial intelligence, everything is written. So it's gonna take a long time to really see what the, uh, conclusions in this area are and the laws are passed.
I want to go to the subject and discuss, now that we've gone to OpenAI, discuss, uh, the appearance, uh, the new model. And I noticed, in particular, that, uh, I don't know, there's gonna be a new model, uh, 4o, ChatGPT, in different ways, for every human being in the world, right? But I have a workspace version of ChatGPT, and I see exactly part of the function, I see a part of the function. What I saw was a better speed of work there. But, uh, with ChatGPT, what's the problem in the last six months? I think he's sometimes very stupider than he was a year ago, and sometimes, like, an adequate thing comes out. And I've already stopped understanding, uh, at what point of time the new model works, what old model works, at some point of time they're testing something, at which point of time, something glues or not. But judging by the funky thing they showed that you could, uh, walk with video and video in real time, artificial intelligence in real time, as it could see eye and help in the world. different things. It's a great case. He's a great case. They've been making videos with a company that helps blind people. Just some space case for blind people, because it helps them cook, helps them clean up, there, not mix things up, how they put in a washing machine. There was a fabulous example of a blind girl saying, "I'm not gonna get confused and put it there, orange shirt with white." Because she's sitting on the phone, and she's got a phone call in real time, all of it. Or she's on food and says I have a bunch of food in my plate, she's got a bunch of food in there, didn't make a mistake, did I not put what a man can't eat. Well, shorter, uh, these things are just unreal.
I'm not talking in a normal life anymore, yes, when we're letting it down, it's not a picture to do, it's not a file to send, it's in real time, and this is just a little unbelievable.
D-Yea, as they have the possibility of this re-real time processing input, on the idea, it'll replace a bunch of apps a-layings as Limitless, uh, Re-rewind, right? So, this is the type of application she's gonna be, uh, a substitute and she might be a real assistant. I'm waiting.
I really liked the part where they actually translated into different languages, that is, right in the dialogue. I think it's new, too, right, damn it?
When I see it, I always get a little bit of a start-up rave, for the makers of the start-ups, because like you do, do, and then do, do that technology, you know, there's a OpenAI, uh, Anthropic or Someone else is coming, and it's all from the box that's getting out. Some time ago, uh, I was wondering, uh, how about using, uh, ChatGPT and the geniuses in the whole, uh, English research networks. Ia had some interesting ideas. Even the company was built there more without me, they're making this product. But I understand that the current update, uh, will do more to study English with neurosets than the idea that was put on the start, right? The same applies to companies that work with videos. Or these are the pieces that were created, yes, uh, so, uh, Sas's reminded of them, too. We talked about them a few episodes ago. It's all, too, in general, already replaced if it gets available. Yeah, again, there's a story we've shown, and there's one that can touch. At least at the time of the recording, uh, access to all of these shown functions is not available.
But it feels like, yes, that if there is, the too is a lot of things on the market.
That's a good idea. And I remember when ChatGPT came and there was a huge number of start-ups, and a lot of people who discussed what would do the same were misunderstanding that, uh, some spaceship that would move, It's gonna be a lot hard not to just stop, but the one you-- you don't, uh, even compete. That's the example. Now there's a function in ChatGPT, I've literally just found it. I don't know when she came. Uh, GPT chat can be said to memorize things. So they had some pre-conditions to keep. I used to keep it there all the time that I lived in this town, and I need to answer that language. And that was an assumption. Now, GPT chat in any chat room, you can say, uh, "Remember the information." Do everyone know, yes, is it or not?
They seem to have added that function.
Oh, can I say-- oh, I saw it, right? So I just don't know what updating it is. I'm gonna have to go. So you say, "Remember," it reminds me, uh, of the information. You're talking about something different chat about, you say, "Chatter, I've got to use it all the time. Remember that. " Then: “Remember”. And there's a whole list of your memories, and you can sort of regulate and build them. Obviously, it's all very childish now, but 100%, 100%, it's gonna be possible to build it very detailed in the future. I, uh, see a very strong breakthrough in the LLM era and in Generative AI when chatbotes start asking people themselves. Not when we ask and get an answer, aaa when you're really gonna have a chat-bot communication with you, further clarifying some details. It's missing now, but once it happens, of course, a huge amount of sophth. That's what, Ilnar, you speak from a lot of start-ups, but it's just not-- it's not the essence of the start-ups, it's the fact that the world will change from the point of view of the inside. Do you need me to go-- there's a CRM system, right? Do you need a searcher, uh, alone? Does that mean that the extra system, there, is that some kind of external photo treatment? Do you even need it? Because, uh, ChatGPT or that system, it could become a non---a distinct operating system. And, uh, Sash, what you're saying from the point of view of the desctop of the application is basically a certain appearance of your, uh, that's an intermediate, perhaps a step. I think they should have an operating system. I think so, huh? Here, uh, but this is a movement on this side, very much on this side.
And by the way, the news is, uh, well, not the news. Apple made a presentation, uh, iPad a new week ago. I'm a stubborn, so, uh, Apple's equipment, and I think that, uh, that Apple will be on the market for AI, you know, a very strong player, and I'm in a personal sense, that's where I have to be with the idea of what it's like. He's got a chip, right? They have absolutely incredible chips. Uh, and, uh, it's supposed to be the news that OpenAI will be coming to its own, uh, different systems in. Different, huh? Nobody knows what. Introduce different systems for use. So, uh, they're really gonna introduce them or are they gonna let their own out? What if they put in the system, they'll release their own in the future? Uh, because they already have an operating system, they have devices, they have chips, they have a lot of different devices, including, uh, even glasses. Well, it's super, super-block technology just, uh, break-up technology from sensors, sensors, chips and everything. Are they gonna do their own thing or not?
Apple released the OpenELM model. I'm just saying that, that's what happened about a month ago, but I think that news is a little in the shadow of their over--the--long negotiations with Google and OpenAI. And I think, yes, here, uh, I think it's really, uh, really, really, really interesting, that, uh, how the OpenAI could learn to play, like, in the big models. So they're a-- they're clearly capable of rocking models, spending and efficiently spending hundreds of millions, billions of dollars on modeling. But they also learned how to optimize models. So that's what Taco-like was never the usual. Usually, for a lot of startups, it seemed like it was some kind of hammer, right? Like a way of differentiating. What a start-up can do, uh-a-a, a concrete case, do, um, cheaper, more efficient and easier, right? And so, in principle, always-- and so now, the start-ups are being created. So they started with the little skirt they covered, then tried to grow up while they were growing up. Noo's like it's like there's no such thing as these models. I mean, OpenAI showed that they're very effective, so, uh, they can... they can really start models like GPT-4o. And the same thing is Google. Google released a new LLM. It's a little-- a little in the shadows, but Google released a very strong little LLM, too. So both of these companies, they, uh, they're-they're gonna be superb, in terms of both practice and big models and little ones. So, uh, yeah, I agree that-- we talked about it earlier.
As long as the start-ups don't understand how to compete.
So you're saying Apple, uh, Apple, uh, you think it's gonna, uh, move in your own model anyway, right? I mean, I don't know if Apple will enter into long-term cooperation with OpenAI, considering that OpenAI is in long-term collaboration with Microsoft or Google, considering that it's theirs. Direct competition. Well, I don't have a-- I can't imagine.
Yeah, you-you're right, so especially right now, just in the middle of a-- I was listening to Sam Altman's sub-categist. Sam Althman and everyone say there's a need for new in-- new interfaces. I mean, uh, replace a physical motto, the phone is complicated, but the interface can be changed according to LLM. And then the question is, what does Apple think? If Apple thinks that, yes, there's a new LLM interface, it's important that they take down this LLM. If they think it's just like a box on the main screen, yes, which might take a quarter of the screen, but it's just a shortcut, then it doesn't make sense for them to do theirs. I'm a w-- the question is how and how much Apple sees priority. I think that's the question.
Well, that's interesting, by the way. They said it for a reason, didn't they? So they said we-- we're gonna have a new chip in the iPad, which, uh, would help launch the AI-conditions--
Yes
...sophth, right? I mean, that's what you're saying. So they do-- they can say, "You're LLM, LLM, but the question is, what equipment is it gonna be? Will you be able to deploy your own LLM on your mobile device? "
Yeah, and that's 100%. And there's just a question here, like how competition is arranged, right? I mean, even if Apple doesn't think they should be like smashing LLM adoption forward, and they don't seem to think so. They, I understand, still have, uh, this very serious competition from other phone producers. Because if some other cell manufacturer has something at ChatGPT levels and it's working on the phone, it's sure Apple-- it can really stiff Apple's position. And here, um, I think Apple is more preventive, if that's what you're saying, right? They say, "We still have the coolest chips, we still have the coolest mobile phone chips. No one can walk us through.” I think they play like this.
Well, it's really like a chip.
And actually, de-- and actually-
Or not?
...in China, no, in the sense that everyone understands that NVIDIA has the coolest chips, but, uh, Grok received some incredible performance.
It's a remarkable performance, uh, inference, uh, Lamas third of 8 billion parameters, yes.
So, very, very, very, very, very, very, very, very, very, very, very, very difficult to say.
I mean, now, there are, uh, different directions, producers. There's a Google, uh, that's a company, uh, search, you know, with a lot of other resources. There is Microsoft in terms of the base entrance, there with the operating system, their collaboration with the OpenAI. They have Bing, uh, their own. There's an OpenAI that goes on a separate... It creates something completely new andy, some operating system, infrastructure, anything, yeah. We don't know, we don't know what it is. And he says, "We'll start making chips." There's a NVIDIA that makes chips, but at the same time wants to enter the market, cloudy... and the market is cloudy. Get in, uh, computers. That was information. They're gonna get into computers, like Dell, er, their own, too, a soup, infrastructure, together with all the others, yeah, create a splinter, an infrastruct... Apple is still separate. So every-- big giants come from different sides absolutely. But, uh, know-how, uh-uh, what the OpenAI does, of course, uh, bye, bye, bye, uh, bye, I don't know, it's just a super-lider for me. Oh, there are many, for example, happy. So, Sasha, what do you think about, like, the companies like Perplexity that did it, yeah, that--that's how you got something else to do? I don't believe that much. Well, um, I kind of... I have a feeling of my own, because there's no base, uh, infrastructure and capabilities, uh, billions, and even staff, infrastructure, like Google, Yes, or like Microsoft. Well, they don't have that. Or Apple or NVIDIA.
Yeah, that's a good point. First of all, I guess it's starting with the OpenAI. So they make announcements that day, for like--- for 24 hours before Google's announcement. So first February, they snuffled Sora before Google announced Gemini and lost their nose. While Gemini 1.5 is a really cool thing, really better in a video understanding, they lost their noses to Sora. Now Google would have put his assistant out again, who speaks emotionally better than before, but still, GP-- OpenAI loses their noses, yes, his announcement GPT-4o. It's really interesting to see. Clearly, OpenAI is putting up a lot of things to make them feel like they're all ahead of us. And I'm actually a little, like, Sa-sash, in keeping with your p-p-- words you said, I'm pretty embarrassed that nobody's ever been in any way controlling the quality of the Open-AI model. So everyone knows it's falling. And it wasn't usually the way to do it in the soft. So, usually, soft-foods, they, well, as usual, they just got better and better, yeah. I mean, uh...
Yeah.
And in general, we know a lot of examples when, because of, there, Evernote, for example, it was, like, great, yeah, like, a note-taking app, but they didn't get as fast as-- like, Notion, yeah, which they did. replaced in a way and so on. I'm surprised that, that is, everything was generally on the way to improve, and in the case of OpenAI, it sometimes went down the path of deterioration, and I think it's not light. It's hard to say about Perplexity.
Well, that's the search, that's the kind of thing that is, like a poundaman-- that's some industry that has a fundamental problem, that LLM is too expensive to use, in blunt words. I mean, searching is one of them. I mean, starting a big model to generate a response is expensive. Iia usually, when you have some business models where you're at the moment, where you don't have a little bit of a little bit of a little bit of a little bit of a little bit of a little bit of a skill, so many companies can put it down, in a brute way, by the investor. money, yes, in a rude way. So they can, uh, bury an in-- like investor money in inference and without a positive youth economy, hoping that it'll come to a close. I think that's how Uber did it for a long time. And I think it's Perplexity for a... the same rate as-- it's the same rate as Uber, yeah. I mean, maybe the youth economy will come down and it's gonna be a business, because they have revenue in the range of 10 to 20 million and 3 billion. All right, yeah, that's the way it is. And it doesn't look like a young man-saving can get away with it., and in case-- compared to Uber, of course, there's a much higher risk in the case of Perplexity, because in the case of Uber, we all knew that no big company would do it. We knew Google was de-- well, Google didn't do it, there, his Uber, yeah. We knew Apple didn't do his Uber. And in the case of, uh, search, in the case of Perplexity, well, it's a fact that--
Of course! Sure.
...Google does. I mean, Uber was about to be a big company for 100 billion, but they had to increase their offer significantly. They're not just, they're taxy, yes, they're also delivering, and so on. In the case of, uh, Perplexity, the risks are high, R-R-F-I think they're much bigger. I mean, I've been in touch with a few investors. Why is Perplexity so assessed, huh? I mean, there's two explanations. First, there's no alternative startup where we invest in search. There's just no one. Aaaa second-- accordingly, everyone needs to invest in Perplexity because, well, often the investor weight-- they want to invest in every vertical. And second, uh, that the market is huge, unconscious, and it is therefore just hoped that Perplexity will solve and at least a percentage of the search market or, there, half a cent.
And that's a lot of it.
Then they're integrating with someone, they're working.
Ah.
Well, it's just a very strange-- very strange story. Perplexity's all functional... First, Perplexity is built and uses side-by-side models. Ah-ah-ah-ah-ah-ah, from the point of view of the one they had in their service, it's like Google now. Google put out the peddate, we all go in, get in, and the Google is responding with AI, right? Very quickly. But it's obvious that it would-- it'll be done. Look, it's obvious that in Google, all the browsers will add everything that is possible from AI to request, and gradually search, uh, will have a huge function-- mostly. I was operating some sort of, yes, or, um, modern, sort of search we haven't seen yet. Some mix of everything. It's obvious anyway.
Yeah, 100%. That's how big companies wear, too, right? So if the company is big and still doesn't see a big risk, it's always better for the start-up to be run. So, like any big company has, uh, some inertia, it's very difficult to make decisions, right? A-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a- But Mr. Vice President will not draw designs, defan, how it should work. He-- it's easier to say, "Do as Perplexity, right? And then it's done very quickly. It's just that, yeah, that's the point, too, to understand what's in big companies down there, that's a lot more effective downwards, and that's a lot easier to say, "You're gonna do Perplexity for all the sp--" Well, you know, maybe, you know, Google can actually find out, because they have Androids that go to Perplexity. And it's simple for the executive to get a request, you get a system, which, on these requests, will be much better with the same design. Here. And with a sort of phrase like that, Google will be done. I mean, but it's hard to start from scratch.
I mean, for big companies, uh-uh, it's easier to cop up more, actually.
Well, that suggests that many of them, uh, actually, and Apple are probably incredibly working, developing, especially given the security of Apple. We just don't even know about it. And at some point in time... And apparently Apple won't do what OpenAI did in terms of product quality. When the product is produced, quality is deteriorating. Apple specializes in producing, uh, the product that works.
There's another remark that this new model is starting, uh, sexing, so much, let's just say. I think the s-- in the Apple case, in fact, it's clear that ChatGPT levels are still not allowed to start on mobile phones. And the moment for Apple will be when it's possible. Ia, but we probably won't know about that before. I mean, they probably have a sales c-c-c- for hardware and software in parallel. In hardware, this irration is about six months at a minimum. Mm-hmm, but it's usually a year. And they've got a plan, there, for the second-- for the 25th year, what LLM would look like and what a b-b-b-b-b-hook for this LLM. And the model will be very small. I mean, it's gonna be a billion parameters or something that's gonna be effective, uh, to get a parrot-- effectively on the phone, right? Because we're even just c-- even just RAM uuu iPhone 15 is six GBs, right? So, well, models should be small enough.
Well, you mean, you're on a phone without the Internet?
I think it's gonna be ah moment for Apple. So, when the ChatGPT 3.5 M-models can start real time on the mo-- well, they can start real time on the mobile phone, it's gonna be like a push that, um, mobile. Interfaces, mobile phones can change and... But we won't know about it in advance. That's right. I mean, uh-
And it could be smooth. I mean, for example, if the conditional Apple from OpenAI is now in place, all requests can first be left, for example in OpenAI, and in time the simplest can first be sewn on local models, and then, I mean, with, uh, growing productivity architecture, uh, chips, more and more challenges can go local. Because I think we're waiting for devices, uh-a-a-a smart, like Claude or ChatGPT. What, like, is there Siri, Alice, Alexa, still remains at the same level they were on parole a few years ago. Aaah, now with a demonstration that was like, this next logical step is to keep it on the phones, but it was also possible on the various devices, uh, uh, those services. use it. Well, here's the desktop app we were talking about, uh, integrating naive into phones, in columns, some other devices into cars the same. I mean, in general, it feels that the next logical step is gonna be like this, well, my opinion, for companies, it's also an increase in market share, because anyway, uh, chat-bots use those who go on the websites, They use web versions. Even the mobile version of OpenAI has only an iPhone, it's not on Android. And therefore, if the list of devices and customer contact opportunities were expanded, it would be a good step to increase the market share. Well, again, I think.
I wonder what we've been discussing the history of what it looks like and what it's gonna be, uh, a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the-the There is a big topic that is discussed at all times in the market, in the market of this, a-a-a-a-a-a-human discussion of AI.
That's what a s-- there's a universal basic income, and it's a conditionality to do artificial intelligence. People won't, uh, work, and people will be paid some money. And just the head of OpenAI Sam Althman, uh, said, uh, in some interview, that, uh, universal income would be about money, but that a man would be given a share of the money. A number of power-powered to use artificial intelligence. And it's interesting, um, the thing is, if you follow the whole, uh, chain and coinization, and business is all. So, Sasha, when you were talking about Perplexity and how much money they generate, in fact, they're not just lost, they're not just not just their model, they're never gonna make it, she's there, she can't. Come on, come on. And especially if we say that a man will be constrained and given a certain amount of capacity. I mean, if I use ChatGPT today, we'll call a number of 999,999,000, and some man uses two, we're having an uneven possession, uh, infrastructure, though I'm free, I don't pay anything for that. It's $30 a month or what, $25 a month in Workspace, right? And if power is restricted to a person and is given limited, and restricted to everyone or restricted for money or restricted for power. This story, of course, is completely changing everything that's happening. So we've been discussing, like, a topic from the beginning about someone stealing some data, and who's actually gonna have access to it? Because who has access? Because in OpenAI, and everything-- and in Microsoft, and Apple, and everywhere they have models that are individually personalized for a number of people, for a number of projects, not resource-limited. And I'm sure there's a level of them, uh, interaction, it's just a little higher. Access level, yes, when you can lift the ban on certain data to yourself or increase the amount of information processed, fill the file in not four pages, but the machine just sticks the book in there, and you, So any resource, uh, there's gonna be some explanation. That's, of course, uh, the subject is serious, and she changes, of course, the world order. Again, she changes the world order in terms of power in the world very much.
Well, that's my idea. Maybe, uh, I was out of the way about Sam Altman's statement, but still.
I've been reading here recently, and I'm not gonna remember what book, an interesting thought that matches your, uh, that the value of the information used to be, that's, access to information. With the appearance of the Internet, especially now ChatGPT, the information is disproportional. So virtually any person can, at any point in the world, obtain some information, as it would for all. So you can't just, uh, get some money out there for the information. And now we're at a time when the great value has time, right? I mean, how much time you can spend, delegating, you're, like, shortening the time. And it turns out that the next stage is not information, not time, but it's probably what you're talking about. Well, we're going there. And it's not that people there were, like, impounding, I don't know, there's a grain on meat, right? Then the money came, then the information came, now the time, and now the production capacity. Well, I mean, somehow, I think it's all logically moving a step by step.
Yeah. Now, in fact, all interested people have free access to GPT-3.5 at least. And after the anon, they say that GPT-4o is also free. Yeah, with some kind of restrictions, which is, uh, a lot of requests over a certain period of time, and then there's a checker up there. What, uh, says or suggests Althman can be added to another conclusion that there will be a limit, for example, of 500 requests per month and no more. Yeah, and you can use these 500 queries yourself or hand them over to someone, sell, e-e-e-h, borrow or something. It's like, uh, it looks like this is a focus, well, being locked in Compute power, yeah, uh, computing power, and it's gonna have to be divided, cut and distributed differently to all. And then, in fact, people will use it. And here's the thing, Sasha, that you said that not everyone, uh, and now it's probably the same access models. I'm very interested in what GPT model the same Althman uses when he says so inspired that there's a dextop-version there, uh, that's really helping him work and so on.
Either he uses the version that we do. But that's what we'll never know. But I'm curious.
I'm-I'm-I'm-I'm here, uh, bo-boo like this grumpy, right? And I'll tell you what, Sam, he seems to be a little bit exaggerating, like, this value from like a GPT chat and a, uh, how important it is, how important Compute is. I mean, for example, I'm telling you honestly, how many LLMs are good for some specific tasks, but in general, there's still a large number of such a flourish routine work that is nothing. I'm not automated. I think while he, uh, slightly exaggerates the value of GPT chatting for a daily man, honestly, well, that's for a regular user, for a regular office worker, I think Sam is so. Exaggerates the relevance of the chat GPT. Here's the chat, so maybe GPT-4o, so if she's already, like, working on my computer, and she's gonna have a dextop app, she's gonna, uh, have a bigger context about everything I'm doing, It's gonna be a real assistant, then of course there's a huge value. But for now, I think it's far enough.
It's me-I'm a , I, uh, listen, to see what she's doing, it seems to me that, for now, it's definitely not a "smuggling" kind of far, but it's like a movement, right? At least it's getting...
But it's where I really want him to, if he sees it, he's actually working local or closed, right? Didn't send Sam over there any information that, uh, what I do at all for the computer.
Look, you're already sending a lot of information, uh, everybody's full. And that's always a question, uh... It is a question, again, er, to use or not to use any tools, uh, and to limit itself because it is possible to lose or not to lose data. I mean, if I'm sending my analysis to GPT chat, medical, and I'm asking you to uncrypt it, and I'm gonna be clear, it's a matter of my sense. Do I care if these data are out or out? Well, in fact, most of my own, I just don't care. Let the world see my blood samples. It's nothing to do with what other data I keep sending in my mail, in correspondence, in WhatsApp, in Telegram, in Notion and everywhere.
And now, during this conversation. It's just everywhere. I got, uh, the other day, uh, updates.
Uh, the other day, uh, updated the phone, and... oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, oh, The correspondence is 10 years old. There's a hygabyte, I think there's 40 or 25. I'm gonna say that I'm gonna be there for you. And, of course, it was so strange inside the feeling. I just didn't get the password, though I kept it. I'm surprised I don't always keep these things, but it was preserved, and everything, nothing came. Then he started to carry the watch, then, see-- the next password can be introduced in nine hours. Well, I realized it was a flexibilities. Oh, and that's when you lose that amount of data, on the one hand, it's a strange feeling. It feels like you've lost almost your whole life. And then you're sober and you know nothing happened at all. I mean, you just want to live depending on and feel that there's some crazy amount of data you're like at any time you can come back. So I'm saying that I'd rather keep all of this correspondence in some LLM-ca, and I'd re-establish it now, even if her other people saw, uh, a piece of that correspondence. Oh, by the end of the meeting what the-- the topic was about ChatGPT. I got a report from my partner here this morning, uh, I got a story from ChatGPT, based on, uh, my article is big. I mean, I have a big article, there's six pages A4 or 10 pages A4. It's a transcendation of my video, and she asked ChatGPT to make a page. The impt is very strict, and it is not possible to add another lexico, but it is not possible to use any side thoughts. So just a summary, very tightly limited, huh? We've been there to check out a lot of that prompt to work. And she sent me today and said, "Alexandre, and, uh, there was a ChatGPT added some more examples. Examples for some items." So there was a comment on how a person could develop, in which directions. It's a professional development, a personal, and it doesn't matter. And then he added examples. I'm so... She says, "It's been a good thing, I think." I'm just here, and I'm reading about my part, which my summary is very, very limited. And I'm reading the first example. I'm looking at a good example. Then I read the second example and I realize that the examples are not mine. I mean, I wouldn't give such examples. They're kind of similar. Tanya said that the information was worthless. They're kind of similar, but they're not mine. I mean, it's not, uh, it's definitely mine-- well, I wouldn't bring it in this context. I said to completely remove it, I don't even want to deal with it and read it, do I? I don't want to get it out and read. I mean, it's important to keep, uh-oh, the identity in my content. I don't have a problem with the volume of content. And, uh, of course, we're in the era, uh, in a big era. That's gonna be a crazy problem, of course. Uh, that's... These are people, especially employees in the company, so the Hippi staff will try to create an endless amount of content with artificial intelligence without understanding any small nuances, yes, I'm not sure what you're talking about. That's how you, Tanya, in the interior design, it's definitely happening, right? I don't know what... I sent her curtains yesterday and said, "I can't, I can't. I have two rooms in my office, two spaces. And, uh, I'm in the same space, I'm saying I can't be there, I need to do something about it." I sent her curtains, I said, "Stacks are coming or not?" Well, that's not the question of artificial intelligence, that's the question of Tana.
I wanted to hear Tanina's emotional, emotional attitude.
So the point is, it's just a little bit of a thought that you-- information on your own, how to hang your curtains and how to choose them, there's a Google, ChatGPT, in books, anywhere. And that's how you use that information, which is already valuable. You'll get through yourself and use it. Like I'm a designer, right? That's a good thing, but it's time and, accordingly, time is worth, uh, money.
Thank you so much. So, uh, we're gonna be out on the canal once a week, and we're probably gonna have a new day off on Sunday. We've always been on Friday before. We had a new time on Sunday. I'll see you in a week. We're always four: Alexander, Tatiana, Elnar, Alexander., technological news, Silicon Valley sites are a lot. Three of us are in the Silicon Valley. In general, we are choosing different interesting technologies around the world. See you again! You'll have to sign up for the canal.