Hello, everybody! We're on ToTheMoon. Technological news, Silicon Valley sites around the world. We're going out every Sunday. OpenAI is redirected to audio-based devices. The explosion of the chat-bot audience of Elon Musk. A re-checking of artificial intelligence from MIT. Stop fearing the Dumpster AI's content from St. Nadella. Homey Robot from Hyundai and Google. Motorola showed AI Pin, a mini-tick with a camera and a microphone that keeps listening. ChatGPT has accomplished 14 of the 48 tasks.
Sas, there's more, uh, interesting news. Maybe even it'll be useful for our audience. OpenAI is starting a Groove recruitment. There, uh, uh, well, even if you don't have a start-up or a start-up experience, but you have some kind of wish and even if there's no idea, you can file an application. OpenAI will select different participants. It even indicates that they will look at the widest possible range of actors. It's true that only 15 people are recruited, but it's not a reason to apply and try luck. Like things happen. The larine card was cancelled, but the OpenAI programme was set to 15 people. Uh, I'm saying there's a deadline for filing, I think, until 12 January. Our graduation is the eleventh, so those who watch day-to-day can try, uh, to get on with this program. The only thing that matters is, uh-oh, that in the first and last week, I think it's the end of January and the end of February, there's gonna be a need for a separate presence in San Francisco. Mm-hmm, see if you have the opportunity to come and visit this event. You can try to apply if you have the opportunity. It looks really interesting. There's supposed to be some kind of mentoring from, uh-oh, those staff, uh-a, OpenAI on the generation of ideas, technical implementation, for some reason. In fact, maybe they'll do some interesting show where you can monitor progress, and then, there's some kind of start-up, ideas, implementation, pitcino, etc. So, there's something interesting going on. And if someone has the opportunity to come and you're in the States or you actually have a visa and a chance to come, apply. Maybe curious.
Yeah. To the word OpenAI. OpenAI, released wait list. Uh, they said they'd have a different new environment, uh, interaction, health. It's called OpenAI Health. Ah, let's see what this is really gonna be and what people's access is. Uh, uh, you can go in on their website and, uh-- well, press the button to wait list. Ah, you're asking, uh, you're doing it professionally or you want to, uh, get some health improvements there. I'm not gonna end up saying, as always, I understand this environmental separation because it looks like, 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-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- By the way, we'll have a special Wednesday with, uh, a man from the Silicon Valley with a strong AI expert who's talking about culture, In OpenAI, why OpenAI does such, not other products. And, in particular, uh, according to his story, it's gonna be that Health is gonna be a funeral, not even a alpha or a beta, but just a little research. But again, um, I don't really know what the OpenAI can bring up to today, except for their main window, which is in constant development. But I don't distribute products like generating pictures or generating images, although there may be some unfamiliar alphas, but still, from the point of view of this basic window, it's just ChatGPT. Those who have already used the Health, write off, as you-- what your impression is. Ah, I think it's gonna be a lot better if they develop, uh, the main window of ChatGPT and it, when you make a health request, help the user. I'm doing a study question, helping the user, not making separate environments, yeah. Or I'm doing a pop--- 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- I'm asking you a question, there, any kind of thing. Business, education, personal development. 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- Or I'm studying something from a scientific perspective, it's inside. It is clear that there are specialized environments, such as software development. She understands that she needs a separate environment, probably. And it's not a fact. In the future, if you look at the future, it's not a fact.
What do you think about this whole thing about what they did again, like, a separate app?
The idea with Health, uh, seems great. Now we're gonna be here for this, but let's see about, uh, separate, separate area inside, GPT chat, yeah, where something else comes up. Now, health. With a great chance, it was Sasha Mashrabov who said it sometime when we were talking about GPT 4.5, that we were just working a lot of parallel teams on different parts of the product. Someone's working there, ChatGPT Education, someone working, that's health, someone works, that's, pictures, someone with video production. And that is when the product, uh, is, in the company ' s opinion, worthy of being able to be ejected into general access, by the words of our next secret guest, to test the product, and then, Maybe it's there, close it, bury it, or maybe move it further. And then, accordingly, this product starts to roll. But with the main line, uh-e-S, the main line that everyone uses, yes, with ChatGPT, it's not connected simply because it's not sure it's still ready that it won't ruin the use of a vast number of user experiences. users. In fact, that's why they're being assigned to some separate part. In fact, in the post card, here, ChatGPT Health, they also raised the issue. They said they'd put it in a separate app, a separate area like this, solely because we need to control, uh, personal data when we work, right there with the medical records, something else, in a separate area, I'm sure this part will be available only inside. You can't get to her outside, so they're, uh, they're sharing this story. I agree it's more like this, well, just different teams worked. Here's ready for the ride. They're rolling, but they're rolling out a little bit. We didn't put it all in the main story. It's about, uh, a separate app. Probably.
Well, if it's not working well, it's easier to bury than to just roll around...
Yeah, but look, they've got inside, we have integration buttons with the side of the epics, yeah. And talking about their isolation, it's very strange, because you can connect your entire calendar or all your correspondence in 15 years to the mail, yes, or connect Google Disk and other different types of correspondence. annexes, yes. Even before you put your apps and their annexes on it. Any environment. The question is, they're also working on a curve and a coso. So they're working, these connections, the curve and the cosot. Okay, I'll get it, synchronize with Apple Health, but he'll get my data. I'm the same way now from Apple Health, I can download a PDF report or I have a-- like Apple Health, which is synchronized with my main health provider, you know, American, yeah. And, accordingly, I'm all synced up. I can unload medical records, well, I can-- I can unload my own-- what they write, they call, there, medical record, uh, information on how you went to a doctor. It's not enough that this is so decentralized in the world in terms of different files, chaos, child-specific old data. It's just, just-- here, actually, too. Let's do this. It's really very, very chaos in medical data, too. And I can load them myself in ChatGPT and with them, do some research on them with the full GPT chat using any versions, there, Pro, anything, no, no restrictions. But they're gonna make me a couple of more requests or a couple of proms. And then I can, while in Health, aaa, connect it separately to mine, like my other chat, yeah. And I can tie it to my usual activity. I mean, but-- and again, Wednesday, we'll have a graduation on this subject. Max is really gonna tell you that very clearly. That's why there's a culture, why some deposits make. I think it's really weird, to be honest. Even in terms of research. By the way, Lenar, you're talking about the test, which is easier to roll, for example. Or they're creating a separate block, but you can't have osabitis built through a separate deposit, and then it's osabitis to come out somewhere, mostly. Well, if you want to create some unique instrument where a man is sitting inside, you're gonna have to use this unique tool to develop. Here. But they're clearly not interested. Not much Max says, Max will talk, and if they ever have an interest in it, it's a separate big question. Here. But this, like this, we'll know about it, we'll know about it in--
We'll know about it, we'll know about it Wednesday. Let's go to the next subject. I think it's a very big story.
Yeah, yeah.
Let's talk about the idea. It's a very nice idea, actually. So you have the possibility of ChatGPT connecting to the devices to be carried, for example, to honey ⁇ there, if you have them in electronic form, they're integrated. They say that the U.S. will be able to connect. It is clear that this is the PDFs that can be wrapped up, but it's like a step forward in integrating ChatGPT into the rest of the world. It's clear that the curve is still working, but the movement is still on this side. I mean, it's a condition, so they added conjectors there for Gmail and Google Disk, there for a calendar, something else. Someone's using, someone doesn't use, but at least the possibility of information from there in the context of chat-bot, it's coming. Now, everyone's been walking there, well, for years now, with the bracelets of Whoop, yes, and you have, respectively, in addition to that, and, whoop, yes, whoop is, whoop is different. Whoop braces go, go with Oura Ring, with any other material that keeps gathering some information about your activities. I was curious, like, yes, what I did. I used ChatGPT to count the calorie. I'm gonna be lying to you, and I'm gonna use the calculator, too. I'm just taking a picture of food and I'm saying, I'm eating it. He's throwing the sizes, the other things. Oh, it would be fun for him to, like, sync up with some more indicators. Like mine, I don't know, the daily weight of the smart weights, mine, there, I don't know, there, the diagnosis of blood, yeah, there, something else. It's clear that all this stuff can be loaded up there and all this stuff in the general analysis is, there, I don't know, its daily weight and some other metrics. I can. But if it's integrated with the futile, the information will be collected, and in the context of that chat-bot that is now answering my question, this information will be very useful. It is clear that the implementation itself is now a skeptical issue. I'm totally in agreement with you here. Yeah, the allocation to a separate area is probably some group of staff who actually leads to it, has brought it to some stage. But the idea itself, the movement itself in the direction is very interesting with integration with the real world. It's like a lot of good to do than what's right now, that's where you can get it.
Well, I wonder if our audiences are writing what you think about and how much worker, working tool. Yeah, it's gonna be good to read different comments. It's a question here, I think Ildar, in the life of the skies, isn't it? That's when someone uses it. Because, for example, if you use weights, they don't get your weights integrated, yes, never.
Not
yet.
If you use it, yes, but it's just a hundred weights service providers. And the question is, how they integrate with them, how they do. Then, for example, if you use it, you measure your own pressure, you're not gonna get involved with them, you're gonna have to hand it over. You're right, it's a single environment, but we also have one and only one, because, for example, there's a Whoop to show.
Whoop is the pressure, Sasha.
Yeah, Whoop's not the pressure. Whoop can-- separate you're putting pressure on and you're putting those data in Whoop. Whoop does not measure the pressure, Tanya.
First time, he's doing it.
You're the one whoop is the one that's the pressure and the but--
I think it's the first time you've been in, and then...
No, no, no, no. He's a pressure-- how can he measure your pressure? How can he do that? It's a separate story when it's separate. He's now taking action by EKG. He's learned to measure. But the point is what? Something might be new. I mean, if we're talking about even the gadgets we're wearing, they're using their different methods of calculating different indicators. And even the HRVs of Whoop or Garmin or Oura Ring, it will be considered to be some of its own, you know, some of its algorithm. Even the blood test, when we're bleeding to standard, I don't know, leukochys, clots, standard numbers, they're differently considered different labs, yes. Not to mention the performance of different electronic fitnesss. And the question is, the interface, the interface to date ChatGPT, while everything they've done, it all looks like a environment you've been doing exactly the same as before, something. Well, someone's gonna make a conctor, and then Whoop will make a conctor. Well, we're not gonna think that Whoop won't, or will oura Ring not make a conctor to ChatGPT? They'll make them a connctor if they're interested in it, too. Because Whoop has his own AI now, as well as all the new modern devices, they come up, which already uses their internal data. Well, half-six-stop, there, curve, and so on. Well, let's see, let's see. Everyone's got it here, yes. I'm putting a bet against this particular interface and that approach. I'll bet against what I'm saying.
Yeah, Tanya, what did you say?
I wonder if there's a point of interest here that, uh, he'll be collecting these big, uh, data, databases on your health, yeah. There, and I think it's gonna end up back in some basic information. You can do sports, eat squirrels, there, carbs in a range like that, well, that's all the information that we know. And when it comes to some serious steps, you can't do it without a doctor. And when you come to the doctor, I was recently with my daughter, who's a year old, I started saying something about ChatGPT recommending that to me, she was so annoying. It's already clear that she's all here in our era. What did ChatGPT say? I mean, she's like, no, you're gonna have to do this, like, like, like, like, like, like, like, like, like, like, like, like, "I'm gonna do it." And basically, without a prescription in America, you're not even a normal drop in your nose, or something. I mean, well, I'm the first one to use as a user, and I'm not sure it's gonna be helpful. Some general information, how I move, eat and eat. Or it's more in-depth, then who's responsible for A, these councils and how to implement them within the existing medical system.
It's about the skies. I was just saying I was going to the pictures. That was very useful to me, by the way. I've ruled out some things, I added some things. I'd be fun if I had integration. There are, uh, sensors that are stuck on the shoulder and glucose measures in real time NFC interface--
Yeah, I had one.
You can actually scan the phone. Yeah, yeah, yeah, yeah. We-- I use them periodically, too. And, uh, if you're gonna stitch, like, the story I ate and the glucose I actually had in my blood, it would be nice for me to do it. I'd understand that there's some kind of product, you know, like I can understand, right? But if it were all synchronized, I got the tests and stuff, it would be very comfortable if, for example, I washed the time when I took a fox of food and when I was bored with the data from it, with the glucmeter and got it. information, how my body responds to any changes in my, there, diet or something. I mean, I still think it's gonna make it easier. I mean, look, I can do this with my hands. I mean, I can scan. I can actually say I ate this at this time. And that's my schedule, there, insulin in blood.
No, wait, you're just saying you ate.
Yeah, I already have it.
No, Elnar is-- look, Elnar, you're bringing a good story. Look, if we sync up all the data that we have in our lives, it's all great. Why would you, uh, make a separate deposit, not the same way you're gonna sync up a bunch of other things in your life that's not about medicine, but you want to compare it? Well, the same thing as my Whoop bracelet, he's comparing me reading night. If I'm pointing out the night, and he's got 30 days, like, 30 days, 20 days a night, and ten days. And he sees loyalty, he sees loyalty. He's telling you, for example, that because you're reading night, you're better asleep, for example. And because you, uh, were late, you were better-- oh, not late, you're sleeping too. And because you two, two drinks a day or not, you didn't have any impact on your dream.
Sas, this is all general information, which makes sense that if you have coffee, you'll be worse off sleeping.
Look again when you have a lot of data. I'm doing a lot of data in person, I'm doing a lot of data on purpose. When you're doing a lot of data, he's in theory that you can find a pattern. And he, for example, writes on some things the law is not found. For example, coffee doesn't see any pattern in terms of your system's condition, as it affects pulses, breathing, oxygen in blood, on, uh, heart rate, etc. There's no legality on the dream, for example. Is there a pattern here for some man? Now, if you'll just put the next thing I can't connect, uh, data, like, from a hundred-- although by the way, from the side bracelets, Elnar, with many, with other applications, the data can also be downloaded, I'm not sure how it is with Apple Health.
Mm-hmm.
How will he analyse them? We understand that ChatGPT is a super-cool system, unique, that can be super-cool. Why would you do a separate interface for this, not inject it? Not to create an interface, some kind of thing that will consume these data and truly make human memory? We don't see that now. I mean, really, what-
It's just a big chance here, Sasha, that's how Sasha Mashrabov said there are different independent teams. They've moved and pushed their project to some stage and they're rolling it out, but they're just rolling it out as a separate project.
Uh, I'm pretty sure you do, yeah. Uh, let's see. Uh, let's go to the next question, to the next topic. Now, 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- Uh, it's gonna happen or not, we don't know. And the models, Max thinks it's not gonna happen, but still. And models for the first audio devices. I mean, I remember, there's a project with Johnny Eve, and it was said that they would release the first gadget in 2027. Uh, now, it's coming, it's information that the OpenAI, uh, plans to probably even release in the 26th year. Although I don't know how they'll do it, but still, yeah. And, uh, industry is obviously moving this way. For example, at the Las Vegas exhibition, a device from the Motorola small, which is available, was shown on CES. Yesterday, my friend bought some pdddata from, uh-Plout, yeah, called, uh, system where you go with some kind of gadget. This bastard writes like I did, remember, buying Limitless. Uh, I don't remember how long it was six months ago, did I? Or a year ago.
More than that.
Yeah, I guess it's more. Limitless, uh, if I'm not mistaken, uh, Limitless, I think they've done some updates, too, yeah. They sold Limitless to someone. If I'm not mistaken, they're selling, right? Yeah. Eldar, don't you remember Mety? Mety, I think.
I'm not gonna tell you what I'm saying.
Yeah. But they had a base failure. So the base is all testing devices. It's clear that the devices will be a lot. I heard a man say, "Yandex will make that device." Yeah, I see the whole world makes these devices. Let's be honest. Xiaomi, Xiaomi and, well, everyone will do that. And Google will do those devices, right? The question is who will make a device that, uh-a, which fundamentally changes something. What's fundamentally gonna change? So, we're going with my buddy yesterday, and he's carrying the phone, and it's got Plout attached. He says, "Why am I attached to it, if I also have a phone, I can also record it through the phone, right? So you're getting an app. You sit there, you think why ChatGPT, uh, won't let that app out just so you can get your records automatically and those records kept there, and then somehow... They may not even be analysed. Let them just get there, huh? Let Eldar say, let him synchronize, let him synchronize with Memories. Uh, uh, uh, uh, uh, this is, like, the Russian application? I don't know. In iPhone, right? Where you're recording the dictophone, huh? Where you're recording, um, dictator, you're recording voice notes on i.Phone. Let it synchronize. I'll have enough of this, it'd be enough at ChatGPT to make further assessments.
Write our viewers who use these devices and what. Because the devices are getting bigger now. Here I am, if this motor vehicle is sold, I'd like to try it because it's like a button and a little. See, again, how it works, how he's gonna have a battery, how often he's gonna need to be loaded. Don't wipe it, Sasha. How much memory he'll have. Yeah, no-- yeah, that's my thing. I washed Limitless twice. So, uh, how much is it that we can wash it? I'll bet, yeah, that's a... that's a... that's a... And most importantly, the most important thing is, what's his battery, right? Because we can ask questions for a long time why Apple watch didn't do the quality of the sophth as Oura Ring or Whoop did, right? Well, um, I don't understand. It's like, yeah, it seems like a space company, and the quality of the soft didn't. But we're definitely in this era. And who hasn't tried, you can read different devices. We may even make a separate issue of this and tell us about the different devices that are available. This is a great big theme. I mean, from the point of view of what you can, uh, especially those people in which there is a different activity, make voice notes and recordings of what's going on, different meetings, their thoughts, Some conclusions, meetings, conversations, family contacts or therapy before meetings and analysis, yes, and analysis devices. It's a very interesting subject. Ah, let's see who's gonna make it this year. I think it's one of the trends. We'll do a separate edition about the trend, too. It's clearly one of the trends this year. Uh, what do you want to say fast? About robots.
Look, the thing is, Google and Hyundai, they're making, 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-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
Boston Dynamics, and then he'll work at Hyundai plants. Well, Hyundai's probably in business too.
Yeah, yeah. But what's the logic anyway? That Hyundai is united with Google in robotics. So they at least announced a partnership to create robots in-- artificial intelligence. And look, uh-oh, Hyundai owns Boston Dynamics? Owner. I'm not really on this subject, uh, I understand. They're gonna do this robot on the Boston Dynamics base, this Google robot. Did you read what Google and I would do on the Boston Dynamics?
Yes, Boston Dynamics, Google DeepMind. And they announced that the first three thousand, or three, robots in Hyundai plants would be working, respectively. Oh, yeah, I didn't know Hyundai owned Boston Dynamics. Well, then, it's getting more transparent.
Yeah, yeah, yeah. But as far as I know, as far as I know, yes, there's more data from time to time that someone buys, sells. As far as I know, they own them. Let's see what's going on. I think they can make a very serious device, by the way. Two very strong companies. Oh, and it's clear that robots are now, so it's not real. Again, if you see CES, everyone's talking about robots. NVIDIA again, Jason Huang said it was a development year, a very strong year of robot development. They, I recall, yes, NVIDIA released special chips for artificial intelligence last year, for robotics. Elon Musk claims that this year will be a very serious movement in his robots. I bought some robot. I don't know if it's coming or not. They're with NVIDIA, they've been testing NVIDIA, right? So Neo Robot, they were testing NVIDIA. He'll show up, he won't show up, let's see. The question is, if the robot is worth $20,000, $30, $40 or 50,000 or 100, you can't buy four or five robots at home. How's that going to happen? I think five-year-olds are gonna sell, there, credit and leasing, and everything. I mean, it's gonna be a salesman, like a vehicle, like, for example, selling it too. I wish I could come to that. Let's see what's really going to happen to them. So how much they can do models to roll inside, huh? How far they'll figure it out. And here, the question is very important to me: how will they sync up with my different data? That's what you, Ilnar, even say, imagine if there was a robot, create a robot that could at least make some conjectors, just conjectors, and he would have aggregated that information. or through it, this information would be aggregated somewhere, and he would be able to really read, study, collect. That would be really cool.
I'd cook you food, so I found my vinchester. Based on your data. A conspiracy. I'm just gonna tell you the case. I had a vincheur with-- I stopped Outlook using professionally in 2011 for a year. And it turns out I've had a data set, 14 years old, a data set, a mail, and they've been in the archive. And I was brought by this Winchester, and I wanted to run this Outlook data set in ChatGPT. So I didn't open this Winchester, okay, there, years, I don't know, ten, twelve, or how much I didn't open, I'm the Outlook, I'm, like, I'm, uh, I'm, uh, I'm, uh, I'm, uh, I'm, uh, I'm, uh, I'm, uh, I'm, I'm, I'm, I'm, I'm, I'm, I'm, I'm, I'm, I'm, I'm, I'm, I'm, uh, uh, I'm, I'm, I'm, I'm, I'm, I'm, I'm, uh, I'm, uh, uh, I'm, I'm, I'm, I'm, uh, I'm, uh, I'm, uh, I'm, uh, uh, I'm, uh, I'm, I'm, I'm, uh, I'm, uh, I'm, I' In particular, for example, in Gmail. Here's the Gmail revolution. Max is talking about it, too. And I got the archive, and one of the main archives, there's three or four of these Outlook archives, one of the files, that's the outluc file with, uh, my own, with the permission, I don't remember what it is, OU, I'm... I don't remember what authorization he weighs, he weighs, well, takes a place, and, pst, he has a pst permit, I weigh four gigabaytes, four points of twelve gigabyte. And I'll notice it's not a modern time, yes, it's been 15 years ago. There was no picture, no photos, no video of that quality, was there? There, I think, was the full-fledged HD at the time. Well, or maybe something. And I think, oh, cool, I'm at ChatGPT now, giving him this file, and he says, "I don't read the files anymore, I don't know, hygigabate," he said to me, or more than 100 megabytes. I don't read the files. I think so, what do you do? I think I'm gonna bury him. I'm starting to archive him, and he's archiving exactly the same amount. I mean, the Autlukov file is already down. I'm gonna put him on Google Disk and feed him on Google Disk with his ChatGPT. So I'm gonna google Disk, give ChatGPT this link. ChatGPT: "Yes, of course." Once I choose this file through the conctor, through the connctor, Illnar, I choose this file, I'm gonna have this file, and after about five minutes, I'm gonna be out of the command line, right?
Well, that's...
This file, right? Well, that's what all my research is done. I haven't moved forward yet. I did, yes. I was, uh, leaning on. I didn't... I wanted a light way, did I? I just wanted him to throw out the Outlook archives so he could tell me who I've been talking to in my life, who I wrote to, what letters. Maybe I wrote some girls a letter, maybe I had some kind of correspondence, which job I had projects, which languages I had. Well, I have a very wide, very big job, yes, professional, and I'd like to raise it, yes. It's like a case--
By the end of the year, the context window will grow by the end of the year, perhaps sufficiently to grow. Try again next year, tell me.
I guess so. Look, from the point of view of the context window, I guess , I'd have been okay if he'd analyzed everything. It's like they give him the mail, like he'll analyze the mail. He's gonna analyze the whole mail. What-- how many letters do you have? Well, if you're a grandma, he'll analyze it. And if you were okay in the mail, how will he analyze it? He had a problem with the hug, see, Ilnar. So he can't eat four gigas. And I'm just making a decision, I don't even know what kind of solution is. I don't want to open Outlook. Yeah.
I see. Well, s-smart, yeah, it's just a part of feeding him. I mean, we need to open this pst-shot file, yeah, it's a part, there, I don't know, a period of some sort of uploading and--
Yeah, but it's a hygigabai!
summary retained. Yeah, yeah, yeah. I mean, it's just a scruple writing, programming, so wait till the context window expands, then you put it all in.
Look, tell me who analyzed his own chat at ChatGPT? They're out, again, Wednesday, watching a special edition. Why am I talking about him? They let us remember a very nice thing. They've released a new year, which means 2025-26. Make conclusions. And we have one of the conclusions we've been telling you who has a lot of chat, yeah, how Tanya has a lot of chat. So you're going in and you're writing, "A can analyze it?" He says, "This was an action, so it's a promo, he's gone." I say, "Can you tell me how many chats I've had?" - I can't. For me to say, you'll export them. " I'm exporting them, and he's... he's hanging, hanging. He can't analyze them because I have a big hug.
Ah, here we are. That's what died. He can't analyze his own chat-- he can't analyze himself.
Yeah, if we go back to robots, I still believe in industrial applications more than in my own personal. Remember when that robot you ordered was just discussed, said it would be good for some hotel network to buy it, because, in general, you have the same rooms, the same set. We have to do something. The circumference is generally not changing. If they're good at it, they're moving from industrial use, in person, yes, when different houses, a much more complex task with, there, different tasks that need to be. I'm not gonna do that. And here's the same story with Boston Dynamics, NVIDIA and Hyundai. I mean, they're forming some sort of state of robots, and they're very cool. I hope we can add videos. Ah, they can themselves replace the food elements within themselves. I mean, there's two power cells on the table. When one comes to an end, the robot can go to the station, replace and, indeed, continue to work. He's kind of tired, he's not resting, he's always working. If a set of tasks plus-minus is determinated, yes, it doesn't change much, I can believe he can take a lot of tasks. Uh, there's a pattern that's 50 kg that can lift, pull up to 2.3 meters and so on. I mean, well, there's a set of tasks that can definitely be given to him. And I believe more that somewhere in the factory they might be acting. You always have to ask, why do you want an atro-anthropomorphic robot, yes, and not these robots, carts, and stuff? Well, it's probably just that we're not gonna replace any of the individual operations, yeah, but what people do.
Interface, if you can express yourself, everything is made under humans. So, accordingly, they collect anthropomorphic robots. I hope that's somehow recovered. There's some interesting news about Caterpillar. They're doing construction equipment. The excavators are likely to have seen the excavator on the top of Cat's signature, yes, that's a cut from Caterpillar right now. Big volumes of equipment. And they also announced that they were starting to build AI-tech with NVIDIA. More to collect data. I mean, the excavator's working somewhere in the field, on the site or somewhere. And all the information is going, going to school. But as a next step, it's obvious that AIka will start repeating the actions that the excavator does, and maybe we'll see some excavators or other heavy equipment there that is, in fact, already I'm not drunk. Also, Caterpillar, for example, has a park, if I'm not mistaken, heavy trucks that work on careers. The drones that are on the quarry are doing the work they need. So in this story, I actually believe a little more than in robots for personal home use.
More simple scenarios of the use of these robots can be imagined.
Look, you see, the question here is the stage--
Yeah, that's a great subject.
This is the scenario. I mean, I'm not interested, for example, in having my robot bed run. Yeah? I don't have one, I don't have one, or there, uh, washing machine unloaded or loaded by me in the plan or dishwasher. In that regard, of course, it's fun if it were all working, and he would go and do just an incredible amount of action and fight there every millimeter. Oh, I have a history of this robot with some, some universal application, yes, when I can teach him to do completely independent things. Well, that's not the same, because the same things, I see, the coolest apartment will clean up the vacuum cleaner that's the pill that's driving to date. As long as you can imagine how much work you need to put in a humanoid robot to clean up the quality of this pill, yes, as you did, and how much money he still owes, how much battery he has to work, how much money he has to do, how much battery he's got to work. He should be driving, doing it, like that. I mean, I see my pills go all day. And what's the problem with the same battery he's got, that vacuum cleaner. By the way, CES showed a new robot, and again the pill that goes up, goes down the stairs. So one of the briefcases that needs to be resolved is a caisal, step down, right? There's more boxes, like washing, like, I don't know, a bladder, right? Or washing, washing the right angles in the bathtub or in the bathroom or washing the toilet. But again, toilets, bathrooms...
There's a Lego and
a wire.
Yes, Lego, wire, yes. Industrial, industrial robots still understand what they're doing. There are robots that are really cool in there cleaning the toilet now. We even showed them to you, by the way. And for me, once again, the robot is somewhere inside, and I have some interaction with him. That's more like "she" than that, there's more things in the black mirror. It's the subject where I can tell you, after the dog doesn't leave or go, so... what's the situation with the kids there or you're gonna be on the kitchen--- you're gonna put something in the kitchen, that's exactly what's inside. No, or take me a bag of land. So totally independent cables that aren't like equipment separate, you know, how to buy it, you know, how it should be, right? There's no equipment that's automatically washed glass. I have one of the most modern vacuum cleaners in there who washs glass, but that's a very specific thing. And we're the only one who uses it in the house. And that's the problem, because when I bought this thing, my wife said, "Oh, well, that's how we got some machine and everything." And it stopped, didn't it? So, this one to make windows, I need some kid to project to keep track of this robot, right? And this robot must still be like a window, not to break, there, not to fall. It needs to be changed all the time, there's water refueling and so on. I mean, here. And our viewers, write to you what's closer, you believe in what year you believe that you'll show up, yes, what cabins would you like to use? That's very interesting to see the boxes of application, isn't it? I mean, when--I was driving here, my buddy says, "What kind of weird people are they? Why do they show the case how to load the washing machine? That's bullshit. Well, when the robot from the basket resets, it means, in the washing machine, in the washing machine, right? I said, "What's the problem?" He said, "Look, I had a girl who, uh, was the hardest thing in her life, it was to put pounds in the closet. So she was very nervous about it, she didn't give it up.
That's a lot of people's stories, by the way. And there's a folding service in America $4 per kilo, I think, if that's what it is. I mean, it's a topic that's been asked. Sasha, you just don't know.
I don't, I don't mind. Look, I mean, I'm just saying that everyone here...
To put it right, you know, underwear, it's got to be colored, and it's got to be distributed by the tissue. It's not that you threw it all in the same pile, and then you threw it. There's, you know, some of your encryption.
But the world can't do that. Dancing, but the world can't do that. I had a lot of, uh, lot of staff who told me-- put their panties and socks on the flowers. Perfectly perfect. There's a lot of stuff, right? But people don't do that. You're getting used to it, of course. If you don't get it, you're like, why didn't I get it? The question is...
Yeah, sometimes you have to be that person.
How much does one-- automatically one for 100 grand, huh? And it's very nice. And here's the wonder if the robot you can teach here is such extra stuff. Because what are you talking about? Which, in fact, or it has to be a specialized device, or a company that makes a robot humanoid, it won't teach that.
She won't teach that. Because we remember, yes, what in OpenAI when an agent is made, they do a case right away, like a sushi in a store to order food to make sushi to their wife or to themselves, there, or anyone else tonight. I mean, here.
Uh, I'm not surprised if the OpenAI releases this case again. It's gonna be very funny. So, with agents. Where, by the way, are the OpenAI agents missing? They're gone. So they're gone. Uh, c-- we'll go to the next one. We have a whole list today, a big list. What you're saying is that Satya Nadell, I think it's a very interesting story. This is Director General Microsoft. He, uh, said that it was not worth overreacting because of the waste content, but it was necessary to focus on the practical implementation of artificial intelligence. Yeah. And I think last time we discussed that too, there's a lot of skill-brained intellect, a lot when there, remember, before everything was there: hallucinations were said, a word. Did you get the wrong letter or something was missing the letter? Uh, not where people, uh, see a superpowered effect from artificial intelligence. Well, Tanya would seem to say, "I don't use GPT," and she sent it, and I don't remember how many, ten or more, or there are thousands of messages, yes, and created, and created more than 1,000 or there, thousands of chat, yes. Here, uh, when you're very involved. And, uh, he's talking about what? What, instead of fear of mistakes or inaccurateities of the generated content, uh, you'll be able to start thinking about how artificial intelligence is to use it every day, that's how to use it every day. And I'm actually talking about the robot when I said it was for me-- and why I ordered a robot. I--I would be very interested, even if some of the cabs were to teach him to do it together. Yeah. Well, the most important thing is that he doesn't hurt me. Here's the big question, so it's today. Oh, Elnar, please tell me about the subject that was math because it's a... z-z-z-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h-h- He's very controversial. M, a title that GPT chat decided 14 of the forty-eight tasks. When you read that headline, it's very provocative in that way, I think Ilon Mac should use it. The Grok has reached an audience of 30 million-- thirty million. They grow very much, X grows very big. But again, when you read that there are 30 million people growing up, you're like, "heh, there's an open AI billion users."
They've been talking about six hundred. He said about six hundred, they're just there for a billion more.
Uh, X, X AI Elon Musk.
No, look, they had a speech, I'll explain it to our viewers. They have actually reached thirty million active monthly audiences. When they put two hundred and fifty or three hundred million users there, they wrote, it sounds like, uh, Grok reached 300 million visits a month. And someone below writes, there are three hundred million users. It's not users, yes, it's three hundred million non-users. It's very important. So the last number of them, their last number is 30 million monthly active users.
I'd be back a little more to Satya Nadell. It's really interesting. Uh, last year, uh, we all heard that AI would start replacing staff, that people don't need to. Even those start-ups are those that are in San Francisco billboards, "Why do you hire people? Take our agents." And accordingly, there's Dario Amodey from Anthropic saying that there's ninety-five percent of the code to write I-ishka, not human, and so on. And that's what we're told that and that they'll get to a level that they'll replace people. Oh, the should add the stuff they print. Yeah, well, we had a word last year and became popular - neurosulp. The slop, I mean, could be like a crap like a garbage that generates artificial intelligence, generates networks. And that kind of thing on the Internet is a huge amount. It's video and pictures and texts. When SEO is generated by neurosets when messages are written with neurosets, posts, articles, etc. And here comes, and, uh, head Microsoft Satya Nadell, he's so little jumped up with saying that nothing AI's gonna replace yet. It's a bicycle for our brain, which is what you use to make your performance more efficient and better work. Don't be afraid of that neuroslop. Yeah, it happens, but you're using it effectively. And you can't replace people, right? That this is a bicycle for your brain to move faster than a replacement for you. And that's interesting because, in fact, the price of companies, uh, probably motivated, that's the high cost of companies, that's why we're creating a product that superrevatory is there, I'm gonna take a look at the people who will replace people and so on. And the head of Microsoft says no, we're not gonna do this, and you're just gonna use it as a copilot, like something that helps you work. I'm not sure how much the idea of Satya Nadell is in this direction. So they somehow rethink more to the copilot and stop pedging this subject, which will replace you. Or... I don't know. Anyway, write in the comments, if you suddenly realized why this turn in the jump was recently taken up by the head of Microsoft.
Here. In the case of, uh, 14 out of the forty-eight tasks, it is worth saying that Epoch AI has math-related benchmarks. One of them is Frontier Math. So, uh, he's got different levels. When the first, uh, wash one, yes, that is, the first level of task, and the Chat-bots couldn't handle it properly. And the people who created this Tier one, Tier two levels of task, said that, well, there's still going to be years before the real mathematical tasks of the model are to be resolved. Tier one, Tier two years ago. Tier three is more complex tasks that are resolved, but decisions, if I remember correctly, have not yet been published. Uh, and that's or Tier's four tasks. So we went to a level where the models started to meet the challenges that were not published online. Yes, that is, they are solved, but, uh, there can be no such thing as, at least, it is assumed that their own decision was on the Internet, somewhere in articles abandoned. Ah, and the model has learned that, accordingly. Speaking of abandoned articles, let me just comment, a little bit of a skilled, but I'll make a comment. It's just that, uh, information was also available. were tasks that seemed unsolved, but they seemed to be unsolved to those people who were a baccalaureate. It was then found that the tasks were completed and published in the most popular journals. It's just that these magazines were not seen by the people who made this baccalaureate. Yeah, well, actually, uh, I think that ChatGPT just found these publications, on the basis of these publications, said that the tasks were done, and then, as a result, the rear number of Baccmarkers was already in place. They found out that the real challenge was already accomplished. And that's how we went to Tier four levels, when we actually have real challenges. It's the front end of science in certain mathematics areas, and there are forty-eight tasks., for example, Gemini has nine tasks from forty-eight, and GPT 5.2 Pro, virtually eliminated it in New Year, has been able to accomplish 14 of the forty-eight tasks. I mean, we're really very complicated, uh, maths are pretty good at it. Yeah, it requires a lot of computing powers. Yeah, it takes a long time, but, accordingly, the tasks are being performed. It's a very good move towards science. And here, we should add, yes, NVIDIA bought a Groq that with Q at the end for the computers. It is likely that , the computers will be present not only in such experimental volumes, but in the industrial volumes that OpenAI companies will buy and those that are taking a long time, They'll be chasing these super high-speed inference computers. Accordingly, it may be worth waiting on, there, Pro versions or, there, on Pro+ versions, that there will be no half an hour of deep research and other stuff, for example, yes, but five minutes. And, accordingly, to deal with some complex mathematical tasks or, there, the programmeistic tasks will be taken up, not for a few hours, but for a few minutes, but it will just cost more. Yeah, that's why we're moving in this direction. Partly, the possibility of models to meet the Tier four levels is also moving in this direction.
Thank you very much, everybody. I'll see you in a week.