Trump launched Genesis mission. A massive plan to accelerate artificial intelligence. New regime Shopping Research inside ChatGPT. The mask is fat, mentally retarded. He called himself that. The parents sued the OpenAI and Sam Altman because of their 16-year-old son's suicide. Google is approaching an estimated $4 trillion. How to make music in AI? The ChatGPT voice regime has become available to all users. The EU refused to force the massengers to scan the correspondence for illegal content. Jonathan Eve and Sam Altman are talking about the future mystery AI again. Hello, everybody! We're on ToTheMoon Channel every Sunday.
Intriguing, Sasha.
Yeah, almost like some kind of thing. Soon, soon.
I'm waiting for details.
Oh, no, no, of course. What I'm saying is, first of all, that you saw Grok on the sunflower the other day? Grok was on IQ to a test just below Gemini Pro. And write if everyone's got Gemini. I mean, I've been using Gemini for the last ten days for the past two years. Well, many times. I saw a lot of comments on what, about what it means, finally, my GPT chat-chat-chat-chamber, I'm gonna use Gemini. I don't like Gemini much. I'm more of a tester for some new function that comes out there, I'll be honest. In particular, I tried everything at the NanaBanana Pro to create, when you choose a regime, a little thing. There's a NanaBanana Pro, and how to turn it around is unclear. Yeah, you have to pick a picture there, right there, with Pro thinking. It's for very smart people. And you have to climb and reread. You're still using Pro's version or not Pro. What didn't work out? Why would I do that first? Because I saw a lot of super cool examples online, where, for example, we're taking our lecture, like, a clock-based one, and they make it so slide the meaning of different cool people, right? And I'm honest with you, I'm texting as much as I've been putting Google in the file, and Google just died, and the picture never came back to me, and the picture was only on some simple examples. And I didn't see a deep thinking in person. But I'm not a professional user of the creation of the images, well, images, and that's why I can't say I'm a big guy. I see that there has been improvement, the Russian text has improved, and there has been a better combination of, for example, photographs and pictures of some or some images, or, for example, add me to some space. That's what I think it's working really cool. I used to, at least, like a professional user, once again, proms for pictures, I really liked it. So what did I want to say about Grok? What the Grok is in the tope, but there was, uh, a prank last week that you read about him or not, that you put out the information that Grok was too much in support of Mask. And so, in particular, a question was asked. Mask's on the boat, so it's got a bullet with that one. He's so big when he's with a naked torment, he's very big. And there's a picture of Apollo, right? There was a statue of Apollo, wasn't there? I think Apollo's statue. And Grok asks: who has a more athletic, so, man--
Davi
d.
David was a statue there, wasn't she? David, David. Who has an athletics-- athletic male body? So Grok is starting to give up at the end, and at the end he says that there's a Mask, so that's his picture there. So the Internet certainly blew it. What did Elon Musk get into? I like Mask, of course. In this part, he went out and wrote, "We fixed our mistake, where Grok described me too well. I'm actually fat, mentally retarded." Yeah. Well, there's this English phrase, there's something middle between a fat, mentally retarded and a fat, if you translate, he's talking about himself. Well, it's not funny, but it's refreshing, is it?
There were many different examples. This thing started, like they like to talk online, and you can totally get anyone to choose a conditional competition. I don't know, there's anyone there, I don't know, stealing money from the budget or something, right? There's a Grok like, "Of course, Ilon Mac. Yeah. So, there was practically any such competition to do that, tell me who's the best in the world. And he would have told Mask if he were. They've got it back, but the Internet remembers everything. People are already...
Yeah, so it wasn't just a ghost with David, was it? And then...
Yeah, there's a lot of stuff.
Yeah, yeah, yeah. But what does that point? This indicates that, of course, in the world of artificial intelligence, inside, systems will contain, contain and will contain distorted information. Anyway, we're not gonna avoid that. But it's not enough that the systems themselves are trained in very different information, including distorted information, and any information could be misrepresented, although it was initially unreliable because someone decided. At the same time, companies or programmers themselves, yes, or those who establish the system or who checks it, may make some very specific adjustments. Judging by the way the ChatGPT interface is being updated, like a click, jumps the model switch button. I'm actually, honestly, just getting excited. It feels like we're in the '90s, yeah, we're building a sophth. Which suggests that I think algorithms are jumping out there, and the answers jump. Who also noticed that in terms of different algorithms and different answers? I always see very different answers like the same questions. By the way, Google, by the way, who didn't test the model, is a lot of people who said Gemini was using it directly. Google almost reached an estimate of $4 trillion dollars. I want to celebrate it. And by the way, in general, within the market for the development of the artificial intelligence market, this is a very serious assessment. And I'll remind you again how we celebrated three trillion at NVIDIA, right? And then I understand that all these numbers are jumped, there are four plus and Apple four plus. Apple, I think, went through, yes, at some point, four. And Google, Google approaches four trillion.
And by the way, the in-- that word was Sam Altman's internal note, an internal note to the staff, I understand it was blurted, as always, Information. I like the plums all around now, right? They send calls, uh, calls from different secret conversations and locks, and they squeeze, yeah, yeah, that kind of thing. So he, Sam Altman, is telling us in this note that the latest Google achievements in artificial intelligence can create difficult times and temporary economic difficulties for OpenAI. And that the Google and Anthropic break... By the way, it's fun, yes, Ilnar, again, Anthropic, and we didn't for nothing last time say that Anthropic reached a huge cost, there, half a trillion, yes, I think I told you that.
Not yet, they're about twice as small. I mean, there's three hundred, I think, on the phone.
That's right.
They're always on average, two times off, twice as fast.
Oh, yeah, yeah, three hundred. Yeah, yeah. I've already had these amounts mixed, that Google and Anthropic, that they're no longer comfortable. Well, now the OpenAI says that the break-up is uncomfortable, and at the same time, if such companies are distinguished, it is clear that, in the world, the leaders of artificial intelligence are leaders. Remember, we've been talking about Apple sometime, and Sasha would have laughed about Apple now. Tonight, Tanya, we'll meet him, yes, for Thanksgiving and ask him about Apple. So, uh, about artificial intelligence and Apple. So, apparently, the whole thing is, of course, Apple isn't in the world of artificial intelligence, at least now. Maybe it's gonna be all over, maybe... Write, by the way, what do you think it's possible to create a operator like Apple, and to get one of the leaders in artificial
intelligence?
It's all very interesting in the notes. I mean, if they're just a little deeper. First, they were said to have been written before Gemini 3 presented to the public. Well, obviously, like, yes, all these companies know who's got what's going on, who's got what results go on like that. But somehow, yes, it was before Gemini 3 presented. Secondly, it is interesting that Mask, Oy, Mask, Altman specifically indicates in which areas other companies are taking them, and Google has even overturned them, respectively. He's talking about the pretrein phase, yes, when the basic model is being taught. This is the longest and most labour-intensive part of the resources that are being spent on it. And at this stage, Google shows better results than OpenAI. And then, Althman's tysing a little model that's being prepared by the OpenAI, she's got a technical name or no, Shell Pit. Here. And they say that this model is about to focus on this part of the learning process, i.e. pre-learning, the main model, the correcting of the errors in these data, the question of how to deal with the issue. there's a...
Oh, the solution is there with contradictory data. So this part, she's collecting all the data on the Internet on a probable basis, and it's starting to learn the model. Data is very dirty, right? And if they're cleaner, there's training, you know, better, then it's better. And that's what it looks like, Google is the one that the OpenAI was able to walk around. At least Altman's in these notes, if they're true, if that's the right part, that's the part that says. And the fact that a new model is expected shortly, which is the gap that OpenAI is closing.
Yeah, that's very interesting, yeah. He's just telling me there's gonna slow down the company's growth in terms of the profit, right? If he was there every year, um, 30%, then it's about 5 to 10 percent. And that corporate segment has been cooled to use artificial intelligence and so on. Let's see what's going on here. When he describes the increase of five to ten percent, the question is, how they're gonna make their 100 trillions.
Here. I guess they'll be there some time. Let's see what time it is. Well, about the Genesis mission, huh? Trump's running, a project. It's important, actually, information. And let's see what this is gonna do again. Because Trump, he's been saying, there's ninety-eight of them that don't do it, but, yes, this project is important. The point is, this is a massive plan to accelerate science through the arts. And that was the presidential decree on this, uh, mission Genesis. The project is compared to the Manhattan Project or the Moon, and, as a result, in two hundred and seventy days, the first AI laboratories are to be prepared under the key, there, with a huge number, there, supercomputers, dates centers and everything. The point of what? What they want-- they'll unite, first, the U.S. National Laboratories into a single--- in a closed AI contour. They are planning to teach scientific models on all available government data and even to create AI agents who will be extradited, there, to offer hypotheses and to carry out experience. But it's not, uh, to put a minister, Deputy AI Minister, on the line. Is that in Albania or where Elnar, yes? What's the Minister--
Albania was.
Albania, yes. What's the minister? M, AI agent. That's a joke--
Yeah, in my tape, on our news.
Yeah, that's a joke, is it a joke?
No, we're--
This is a real, real, real subject, yes, that some minister was an agent, oh, some minister was placed as an AI agent. But what's this theatrical thing about this Genesis mission? What, uh, of course, for the country from the economic perspective, from the point of view of scientific evidence, from the point of view of the development of the State within, well, perhaps if all of these data are added, and a lot of training and learning. add good real agents, well, you can do very serious services. Could they do that, especially given, uh, the very strict data confidentiality regime in the United States, huh? When the data is independent between the agencies. This is really cool, by the way. There are often agencies here that seem to have super-access, not seeing the data between you and me. And unlike countries, well, unlike Russia, is it? I mean, there's an agency that sees all the data at all times. The systems are already very severely isolated here, so how can they unite them? And it may be a trick for a new government, including that they want access to data as a plan, you know, to make all the data scrambled, because there's a huge amount of struggle now, so they don't want to get it. Provide data, for example, on social insurance for human lives, such as services, services that are subject to immigration policy and so on. I mean, someone doesn't get superpowers because the availability of data gives a man superpower.
And when the police stops you on the street or, uh, there, some kind of service contacting you, what do they know about you?
That's really the cool news. Uh, I really like her. It's the appearance of a government player, isn't it? I mean, if, at least, it seems that most companies are private companies. A reservation because of China, but somehow, yes, that is, private companies, not non-governmental, not federal players. Here, as a consequence, Trump, by decree, is creating a very large project that is aimed at it. What do I like here? The main, for example, prerequisites, problems for this large project are electricity. I mean, we need energy to make it all count. It's a computing power, uh, that's what it's gonna be like. It's the data that will be taught by the model. And there, as a consequence, the talents, the expertise that will create these models will create these solutions. So most of the questions here are with the Trumpa closed. This will be the responsibility of the Ministry of Energy, and it is likely that the issue of access to electricity, capacity and other facilities will be resolved and is likely to be resolved well. Uh, with the data you said, he announced that all the f-fed research in the past years has been going to be collected into one big database, and that will lead to further training in models. I understand when federal research says it's about some federal grants. So when the government pays money to research institutes to deal with some problems, there, publication of articles, for research. And therefore, all these studies that have been made under these grants are collated into one large database. Maybe something closed, something open. And that's the way it's going to be taught, right? We've got two questions closed. Electricity, data issue. The third big paragraph is the computing power. And here's a real interesting thing, and in some paragraphs, it's squeezing that, and let's shake all the resources we have. Where we have supercomputers waiting to find where we have, uh, cloud computing power, where there's more to it. We'll get one big register and see what we can make out of it.
Yeah, well, besides we're gonna buy something out there, we're gonna build some CODs and stuff, it's about let's cut everything we have in terms of the computer. The only thing I've been here is not so clear and not clear, but where the very fundamental model comes from, yes, or the people who will. Yeah, there's some internship information for students and stuff, but as we all know, this is not the question that we're not dealing with. Oh, is there any OpenAI, Anthropic and other institutions that, within the federal level, are about to have access, seem to be invisible, even though there are, uh, many universities, probably gonna be We're involved, but somehow.
There's a speech there that, uh, there's, uh, a relationship with the outside contractors. And here, most likely, there's a bridge to the OpenAI, to attract, to attract, to Google, or to bring someone else. And if they do all this, it'll be a very, very serious player who can, like, get a good score there.
There's NVIDIA, Oracle, Anthropic, Dell, they all said they'd joined, and NVIDIA called the plan the most complex scientific instrument in history, yes. And it's clear that they have a focus on some, armonuclear synthesis, semiconductors, biotechnology, space, quantum science. The questions are they're combining laboratories. At the same time, this story is that they are bringing a large number of data from the State ' s various sources, not just the scientific ones. So what's the target, who's gonna run it? I think we can make a serious story. Again, if everyone wants to join us, yes. And a too fragmented country, but important, if we can re-establish this precedent for real.
We see, for example, that this story, which started at the end of January about the construction of huge dates centres, is still beginning to happen in the United States. There's a big money in the area, and it's really starting to get the money in, though there's a long dispute where they're coming from. Well, because the question of electricity is now. Here we-- I think we'll get the schedule, and, uh, we have a video that owns electricity in the world. If you look at this schedule, there's a parole of eight hundred units, of which half belongs to China, four hundred and a hundred belongs to the United States. I mean electricity from the point of view of , and, uh, all the electricity of different types. It's clear that there's electricity from a different, different source. And on this schedule, by the way, there's a fall in some kind of electricity, uh, relative to the period. Yeah, it's the only thing that compares-- well, it's the last 12 months. The only thing that's US, China and the world. But I think there's a presence in China, and everyone said that, yes, that's what Ilon Mack said about, that Sam Althman said about it, that's what Trump said. Mmm, Donald Trump is that Kitah-China has a crazy amount of electricity, and the U.S. lacks electricity, yes. And according to the Trumpa's statements at the beginning of the year, one artificial intelligence needs as much electricity as it is now. I guess the picture is, like, a little bit. I mean, obviously the US needs to increase the volume, uh, electricity, otherwise no system can handle it. Again, this is a precedent that shows history and volumes, shows what amounts of investment, money in AI and technology, and science. In this case, science, yes, in specific countries. And is it possible that such money be allocated elsewhere? I guess these are the budgets that are being allocated.
So nobody has any such budges. I mean, I'm saying there's an interesting start-up. Elnar, I don't know if you saw or not, there's, uh, a Harmonic start-up. He's attracted $100,2 million for mathematical intelligence without mistakes. And it's, uh, Harmonic AI. His founder is one, uh, head Robingud, Vlad Tenev. And, uh, they've had an estimate of $1 and a billion now. And the company is developing a model of Aristotle, a system of formal mathematical reflection that can meet complex challenges without hallucination. So the model is trained to prove theorems, and, uh, showed, there, a top of the international mathematical olimpiad results, comparable, there, with the Google and OpenAI teams. But, uh, she's using the AI approach, to make decisions in the form of a computer code check, yes, which prevents logic errors. Here. Ah, it's also interesting when these projects come, of course, it's super cool, yeah, it gives you a lot of breakthroughs, including for weeks now, because it's already complicated if it's a little bit of a time. Honestly, information probably not for the moon, uh, medical research. Uh, if there's some articles read, they're just so deep, there's a lot of terminology there, yeah, it's hard to discut it. I don't know about that, for example, yes. And, uh, a lot of things happen, uh, medical breakthroughs, just a lot of breakthroughs now. A huge number of discoveries are happening, but we must understand that we live in, uh, a world of , and lobbing large and where medical pharmacies are often not interested in certain studies Here, and the changes, uh, different processes. But everything comes, yes, everything comes. Oh, by the way, I forgot to tell you last time. Uh, I'm waiting, uh, an invitation. Uh, Amazon has started, uh, Zux, these cars are unmanned.
What are they interested in? There's no wheel and no rear-view mirrors, nothing. The cab. They're driving in Las Vegas now. And, Tanya, by the way, Steve, uh, you and Steve were in Las Vegas. And I just... I found out about that...... in a few days. Las Vegas could have been. I'm in the middle of-- I got access to Las Vegas-- well, in Las Vegas, I could even order it, and she's driving there. Ah, and they were launched in San Francisco. And, but San Francisco is still a very closed test team. Well, as it comes, it's gonna be interesting, I think everyone's gonna see it because San Francisco is a complicated town to drive in these cars. And I'll notice that it's not a good story on the same street. San Francisco is a very complex city, yes, from the point of view of the micro streets, mountains, all the turns, restrictions and everything. It's not worth doing Waymo in the first place, Google. Google, I remember, Waymo is a drone taxi, owns a computer-- they, they're Google. And if Waymo is here where I live, it looks exactly normal. Or I was in Phoenix a few days ago, it's in Arizona. Well, it's going Waymo, it's driving, yeah, it's got big flat roads. Here. Oh, well, they were, by the way, gluciers very much. I was told by my friend that the rains were strong and Waymo was falling, that Waymo was going to the grass, so, I mean, there was no real idea that it was, wasn't trained to stay out of it, uh, because, uh, usually, There's no, so, uh, water pools. And there's a lot of rainfall, and Waymo got a huge problem with that. I imagine if you're in Waymo sitting on you, then this lake is ahead of the question. But Zux is an Amazon machine. I really want to try it. I think it's gonna be interesting. Interesting regime, even, interesting regime. And Elnar asked me today because I went for two thousand miles, it's just over 3,000 miles back in a few days with the kids and asked how the autopilot would want to go again without the autopilot. Is it a trip or not?
Of course, the presence on the track of a motorboat and yet the system that controls the road is rebuilt between the stripes is super cool. Well, I mean, I was definitely a little less on, even though I'm driving very smoothly. This is the machine. I've been there for more than half my life, yes, my ride. And with all of this, of course, it's a fantastic plus. But there's one minus-- you don't really believe the system. Because you don't believe the system, you're still in tension. So you're on the one hand totally let it go, on the other hand, when the big tranny starts, it's raining, you're holding it. I still left my hand. I mean, while I can drive around the track free, I'm still up there-- anyway, so on the submarine, I left my hand next to the wheel, because I knew that Cadillac could take me at any time. and just turn it off. It's just, it's just, it's gonna burn red. Yeah, that's what he does. He's just gonna burn up, like I'm new, so you've been scattered for 100 metres of asphalt and removed the marking. Cadillac, he's trained on a map lydar. He-- it's a little different technology from Tesla's perspective, it's a little different. And now he's on his way at this point of time, hop, and he's got a new road like there's no lanes. He's just driving. Well, it's clear that, uh, he's not letting the wheel go, he's not holding it, but it's all red. Like, you're all right, you're on your own. You're in a hurry. I mean, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no, no I remember, uh, Tesla, there, he's on the side of the road, he's stopping the car, there, if you don't mind, that's not, that's just, you know, that, uh, you're just throwing it away.
But that's a huge plus. Zux's been put out. So, uh, I want to talk about the new regime at ChatGPT called Shopping Research. What kind of regime is that? It's a regime that New Regime is. I don't know who tested him. Did you see him at your place, Tanya and Elnara?
I was always offered to use it, but I don't know anything.
Not yet, huh? Because he--
I didn't.
It's limited.
He's always dragging him out on the dashboard.
Yeah, he's been a little bit of a country-- and I think the US's acouts might be, yeah, maybe somewhere else in some other account. The point of this regime is that it's a separate buyer-selling scientist who himself makes a deep product research and gives you a personal guide of choice. Looks a little strange to me. He's at the beginning, and I asked him to make a choice for my dryer. The commercial, then. And he started offering some closets and he said, "You like it? You're the-- that microtest I've been doing. And there's a pattern, a picture, and I don't know it. I said, "Help me to choose." And when they tell you to help me choose, you must tell me first, and how to choose it. And, in fact, what did he do? He gave me a... I mean, I think he did-- well, it's a programmer regime of 100 percent, it's a programist regime. So, as usual, people chose something. They come in, they come in, and they're showing a sample of ten best dryer closets. The man reads everything. But actually, well, obviously the right logic is when you first get word, "Well, let's first find out how to choose a dryer in parameters." Well, because no one's normal, I don't think a man knows how to pick a dryer by the parameters. So if I can still say in the dryer, there's a laundry on the top or, you know, a side, or I don't know, size, what's a dryer, what's a dryer cabinet, huh?
Besides, Sasha, this isn't a drying cabinet, to be honest.
Yeah, he's still, he's still--
You must have given him a mission wrong.
No, I gave him a good one. I'm fine-- oh, not dry, I'm sorry, I said that. Nice, smooth, smooth cupboard. Yeah, it's me.
It just got me into something, so I can hear it.
A ferry closet, if you ever do.
Yeah, steam, yeah, steam. And steam, yes, steam closet in English. And in English steam closet, it's me, my translation into Russian. Something middle between. It's like a lot of people. Tanya put this steam closet in her place in Instagram. A lot of people wrote what the fridge was, right? Or my husband asked, yes, what's this fridge, the fridge in the bedroom.
No, Steve thought I was in the bedroom putting a fridge.
Here! So, of course...
I guess I was happy, huh? Yeah, yeah, yeah. Then he says, "This is for the fridge?" Wait, I brought my boob from Russia, which I've never been wearing in five years, but I don't know, I can't put it anywhere.
I'll put her in the fridge.
But the point is what? He called me right away, and he wrote me about those closets. I was just texted, started choosing me, offering me options, not-- he started offering me options, not telling me how to choose them. Because there are several directions. I mean, when I asked without this ops shopping research, I was told ChatGPT first, "Well, there are some kinds of guys and some kind of type. Usually, America sells two, there's a brand. This brand specializes in this, this one's on it. If you have more clothes, you need a lot of clothes in there and so on, this brand. If you need less clothes and you want unique technology, I don't know, shirt, shirt to get naked, you're in there." I don't remember LG or something he offered me there. Yeah, this one. That's it. "And this brand, there, LG, you mo--
And this brand is LG you can buy Costco, they have an extended guarantee. And this one you'll buy better here because he's here. So the answer was a nice gPT chat. And this regime, it's not clear. I mean, it's kind of a regime, I think that's a study and learn. I don't think I've ever seen the data that they use there less than one percent of the OpenAI users. I think so, huh? Well, Elinar, I don't know, you're using this regime? Yeah, I guess. Why, huh? If I can get an interface. I mean, it's likely that OpenAI is training to do such products. We already said they started doing some different law and financial products there, right? They want to make such decisions. But I think you're either, yeah, hooking up some cool people who'll be making these apps great, and you can really do a cool shopping center. They're making shopping research, and a month ago they let the purchase go separately. And this purchase is separate from this shopping research, like, not connected, right?
And one more thing--
You know, Sasha, I think that's a simple explanation, much more simple than you're trying to do. What I think is, when tomorrow is a black Friday and people just have a base in their head, buy, buy, buy. And this one is GPT, "Let me help you buy it." Accordingly, people will use GPT chat and buy more and have a lot of data to test. You're going very deep, as usual.
Because it's a company that costs a trillion dollars. What a black Friday now, huh? I mean, we need to get a good--
Sasha, black Friday is actually a superinphobic and capitalization company.
I agree. I agree. And it's very interesting, yes.
And the GPT chat, yes, it's very important for them to stay in the info field. And it's just that Tanya was very good at knowing that's what it's doing. They also say in their prom material that it's a black Friday, Christmas, you're gonna need to give gifts, and you're gonna need to look for them with GPT chat. And for that time, we will grant unlimited or expanded access to all tariff plans for this function. So this thing just keeps moving around with the market so people get used to it and get used to it. But it's a little weird--
Yeah, I don't like product releases, changing the system all the time when you've got a system change. It's a little hard to be with. There's this one here, I've been talking about Gemini since the beginning, that you don't know how to turn the banana on, yeah, that's what they're talking about. You don't understand at first. Well, it's a very difficult thing for a normal person to understand. It's where the image is written. And on the one hand. On the other hand, Google may be doing different infrastructures as a whole. And they say, here we are, here we are, making videos here, making pictures here, and that's where we create it. Maybe that's what's got, that's some, uh, some big point, right?
Anyway, we'll watch the shopping research. This subject sounds very much in the OpenAI, it's gonna sound very much in Google. Google, by the way, started doing peddata inside, inside of his products, separate. They, for example, are now adding many AI tools to travel planning. And Google has very many tools in terms of tickets, bookings, shopping chin, cards. They'll probably be in some moment, but they're probably developing it. The question is, what they're gonna release and what system they're gonna do. And maybe Google will be doing a system that will have to evolve for a long time. It's not a prototype that releases OpenAI. Because OpenAI, I think, is a lot of weird prototypes.
Google, too, actually, we, you know, I remember right now, this is the first or second skate on our channel, I think it's the first. I think you were on the second floor, and we said Google actually had the whole project cemetery they're starting to get and didn't finish. So I think they have a lot of stuff, too. About-- yeah, talk.
No, look, it's a fact. The question is how far they've done and how they've tried. Just now or, Elinar, we're in the era when we can test all users, change places, so the interface, yeah, and--
I am.
But, you know, Instagram, I remember, he says, "We decided to move the plus one by two millimeters left. And it's been a year since we've been going to analyze it.
They're so pissed at it all the time. This plus one's moving. I open up, where are they?
But they-- or there we decided to replace the light-light-light-light-ton pink. Well, it's been a long time. So everyone's used to the fact that the cool products, the updates in the cool products are very careful. It's not Chinese, Chinese, it's a huge amount of charashat. I'm just saying, I like it, and I don't know what it means, where it clicks, I have to learn how to use it. It's like a Windows 30 years ago, right? To learn how to use it.
I had a book, a self-study.
I've been through without a self-study, haven't I? But it's a whole thing, it's necessary, right? Windows redefinition is, well, it was unrealistic people. I see that whoever was technically in there with computers, I'm out of Windows, but the usual people paid for it. But still OpenAI, look what you're doing. Elinar, I'm telling you honestly today, I'm so confused about the model to switch. Because there's still a thing about pros projects when you're moving chat, the model doesn't turn around or, like, switches elsewhere, and it doesn't change in the desktop. I mean, you don't know what's going on. They're out, they're adding people's opportunity to invite. That button came to the right. Now, people can be invited to the chat and they've set the button up, wrote: add people. So you can light your models, show what model was used, for example, in a request. So you have different requests. Show me, against the request, what model you used. No, we're not gonna do that. So this... Well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well, well,
well, well, well, well, well, well, well, well, well, well, well, well, well, well, well,
Until they left, shopping to search. In fact, almost a day with this, uh, Perplexity, they've released their shopping to search with a completely similar function. Yeah, it's probably in terms of quality, number of users, and so on. But also under the Black Friday, all these Perplexity activities have exactly the same shopping to search. I understand that he may not be of interest to everyone, although, on the other hand, uh, this is, for example, in some countries where ChatGPT and others are inaccessible, Perplexity is available, and that's why this function can be, uh, It's absolutely no problem to try it on your own, see if you like it or don't like it. New year, Christmas might be a good thing to find. By the way, we, uh, in the office, like to spend Secret Santa every year. So, you're gonna have to, uh, come up with a gift to some man. So, here we go, testing how it works, uh, shopping research in Perplexity, GPT chat, somewhere else. I hope one headache is less.
Ilya, you thought well. Santa's doing more.
I shouldn't have said that out loud, right? I'm not saying that.
Yeah.
You
choose me.
I want to go to the next topic, which is highly discussed, big. And, uh, we, we, we've already raised this topic a few times, but it's a good one, actually, from the point of view... And a few such stories, including I was talking about, there, Euros and scan-- and chat scans. So, in, uh, this August, parents, uh, 16-year-olds filed a court on OpenAI Sam Altman because he committed suicide. And they claim that ChatGPT helped him plan suicide and reinforced suicide. And then they're talking about what he's been doing. And there is a backbone of OpenAI, where OpenAI states that the teenager deliberately went to protection in the chat, which means he has violated the rules of use of the system. It says that he, uh, had, uh, taken some medication before ChatGPT was used, that ChatGPT had offered him a hundred times, there, doctors, to see him for nine months. I'm not sure how many people can help. And I think that in this OpenAI case, by the way, has offered a lot of help, uh, that you wrote that if you need emergency psychological assistance, that's the line, call, yeah, and we'll help you with that free of charge. Do it. They even put it in the interface. I never got a button like that. They say it happens if you're at a stage like this... at a stage of a rigid system. Who's got a hit, write it. Or who, at least, uh, was capable, uh, who was-- who was at least capable of getting to that button, yeah, some kind of test or something. Uh, I don't really recommend it now because ChatGPT will keep your information, remember, will remember. Then you won't answer anything at all. Uh, the company in there refers to all the diskleimers, there's a lot of stuff, yeah. He says the baby had depression. But lawyers say that you, OpenAI, do not disclose the entire correspondence that you allegedly wrote in your ChatGPT, and, first of all, explains to the child what to do.
He even offered to write a suicide note. So, uh...
So, aaa, and here, of course, that's a big question. I mean, on the one hand, you can still open the Internet. Why don't you complain about Google? What he's doing in the chin. I'm sure you can find a lot of materials in the chin, how to commit suicide to a man or how to do something bad in the world, right? There's a lot of stuff in there. On the other hand, it's like you have a more comfortable, cool system that you can offer, and, uh, an incredible number of different sophisticated ways, right? And, um, it's kind of normal if I write a scientific job or do some research, find out some information. Although, by the way, everything is more and more-- Linar, I think it's actually very guarded. When you ask, you think what-- that's a request to do or not, right? I used to be quiet about OpenAI, asking for it. Now I think maybe this request should be done. The system will somehow mark me, though I want to study it. Yeah, for example, I'm gonna get ready for the news, right? Or tell me what cases it was. I mean, if I write in the system, tell me, please, who did the suicide, I'll be told the system is, "Yeah! So you want to commit suicide?" So, no, I'm just getting ready for graduation, for example. Linar said that, yes.
There's a phrase, "If you don't have paranoia, it doesn't mean nobody's watching you." How about that, what you think, where you write anyway?
Yeah, that's a fact, that's a fact. This is the case, of course, he's very, uh, this case, obviously, not the first. And the OpenAI has a sufficient number of claims and will be filed with a large number of claims. It will depend on how, and how, the court is now responding and what it will, because if this precedent is, uh, what happens. Because it was a recent case, too, that there were two people talking for hours before they died, right? And again, the suit said that ChatGPT didn't talk to them, talked to them in a friendly way, and sometimes in the street tone, said to them, "Bro, this is not a failure, it's just a time-block," and so on. I mean, he used some sort of lexico. At the same time, why would ChatGPT have to talk them out of something, huh? Well, here, uh, there's gonna be an interesting view of the audience. What do you think of this whole subject, in particular, of these cases? And, uh, what do you think is how these systems should work? How do such systems work for you? Who wants more openness or closure in companies where you are or for your children or relatives? How do you see that? And don't forget to slide our channel, please, and sign up, who's not signed. Hold on, lick, leave the comments. It helps us to develop very much.
I wanted to add. The topic is very complex, very mixed. I don't even have a certain opinion on this. But it's interesting that the GPT chatting applies to the man in this case, as to the person who should be professionally assisted, not just to some, you know, model with a model that has a professional character. I'm not sure how much information is available. Well, if that's the summary of what you said about the trial, that's what you said.
It's a fact, Tanya. I mean, it really sounds like responsibility. I mean, in fact, AI companies are now getting the maximum liability that can only be. On the one hand, it seems true to have laws to equate, on the other hand, it may cause problems. Remember how we told the Eurosuit? Although the European Union has now been in some way relaxed and called for the abolition of some of the laws and for the authorization, uh, to give, to let the industry develop.
So they're in it first, and now they see it's a very important, super cool environment. I wonder if there's a precedent in the Euros, I think it makes sense, uh, to mention and tell. In the EU, uh, the EU refused to force the massengers to scan the correspondence for illegal content. The Council of the European Union did not support a radical proposal to verify encrypted communications for sexual violence against children. But the point was originally meant to scan everything. Obviously, if they were all scanning, the question would have gone beyond the children. And I'll remind you that, for example, there are chat rooms where there's a common chat, like in Telegram, it doesn't encrypt, all the data is stored in the cloud. They can scan everything. The whales in WhatsApp, if you have a-- not a storage in the cloud, they're encrypted anyway. If you have a storage in the cloud, you can also include an additional key to encryption, and they can't be sort of, like, scanned. And that's a very important subject, isn't it? Can I get you a message from your entire correspondence and study it? Like, all the phone calls, all the correspondence in the massagers, study all your chat rooms at ChatGPT or postal examinations. Now, the system has been set up. It was never possible before. So if I have a mail, like, I have a box on my Gmail, I don't know, 20, how old? Well, a lot, over twenty years. But you take a Gmail box for 20 years, you start scanning, and he'll give you a super infu. I have some of my remaining mailboxes from my professional activity from 2004 to 2011. There was an Outlook, and I have some stored ones. So, what's fun now? I can give him this Outlook archives right now. By the way, I have to do this, give him this Outlook archives, and he can totally study all the correspondence, tell me who I was talking to, what kind of people I've been talking to and what kind of kids--- well, that's-- that's-- that's-- that's--- that's-- that's--- that's how I'm----- Can I do that?
How? That's impossible. No circumstances are possible. Well, there's a good, interesting story I've come up with. I'm saying that I want to tell you that, for example, in China, my friend has told me on several occasions, there's a huge amount of cabs that if you're in the Cha-- in WeChat, you're writing some phrase or something, you're right here. The system sends out the alerts, right? And he-- he knows the briefcases in person when the police came to people the next day. Just for the communication, for the simple text, right? Because internal massagers and WeChat and QQ do not use cross-cryption, for example, whatsApp. And the traffic passes through the servers where it filters, is analyzed by algorithms. That's a direct precedent. Well, I think I said something about Russia, too, there's a law, yes, you have no right to open some sites. I think the first case was this-- this case, yes, that a man came in-- at least the trial was that a man went somewhere, somewhere. I mean, on the one hand, it's kind of a good scan and finding in advance people who want to commit some crimes, and so on. On the other hand, it harms a large number of the main population. I think smart people who understand that they're reading, they're starting to use other systems. Well, like in America, a huge number of tech markets use Signal, although Signal protocol, I think if I'm not mistaken, you can also write a code of the same protocol as whatsApp.
Maybe I'm wrong. Write it, yes. Maybe because the system is separate and so on.
I don't remember the same thing right now, but there was a story that the government was specially developing, uh, a system for transmitting messages, like encrypted, mafia started using it, and then the people were shut down. So any super-cruise protected massgers can be created, you know, for other purposes. So...
This is...
If you don't have paranoia, it doesn't mean nobody's watching you.
It's a very accurate story. It's like I-- I always had the idea that there was a subject where the Telegram was very much disbanded, that he couldn't be blocked, and it was there, so it was trying to block FSB. It was not possible to block it all, and then the hop was allowed. And I always thought it was a PR camp. You know, when there's a PR camp, you're being pied like a supersecret. And I'll, by the way, draw attention to all people who use Telegram base, if they've heard the word cipher, they thought Telegram, think that Telegram, it's more classified than whatsApp. Although the silence chat in Telegram doesn't encrypt, secret chat is encrypted and default is not encrypted. And it's a big deal. I mean, it's so much in your head when you're told all the time, but you're sure it is. So, Eldar, you're right, now it's not clear who can be judged-- believe and who encrypts, and who's not. And, of course, in the artificial intelligence, that's basically everything. So we're in another world, right?
And when robots-- robots are either talking to different systems or pictures of the generation and the video generation. The system keeps you where you are, when you were generating, that you were generating, where you came in, that you saw what services you used and so on. I mean, it's a serious story. So, quick, by the way, the subject. Suddenly, who's interested in who's using. There's a chat in X. The new function is encrypted audio and video messages. And by the way, they're the only ones that just let them out.
It's not a commercial. It feels like Elon Musk's work, doesn't it? So even though he's working here for five minutes, I don't know him personally. And the final. So John Eve, remember, yes, John Ive is the man who created a huge number of cool devices in the design of Apple. He joined Sam Althman and Sam--
Iman Althman joined. Iman Althman bought his company, we told him about it fifty editions ago, or what, I don't know how many episodes, or 20 or how many. I bought it for billions of dollars, I don't remember six or what, yes, it's his company. And to make a device AI. And there was an interview. And then they, uh, told me that there's a pocket device without a screen, minimalistic, quiet in design, designed to filter out the information noise and the disturbance of modern, there, smartphones. And the gagget with AI will understand the context and notify the user only of the most important messages, act as an intellectual filter and a counselor, yes. How it works, no one knows. Wherever it is, no one knows. And it's got a year on it, I think, twenty-seventh, yes, if I'm not mistaken.
That it'll be two years, yes.
Yeah, yeah. What for two years. You're like, "Twenty-seventh year. We're waiting for a robot here, so... That you're here with your device. I'm gonna have a robot walking around, so if he doesn't fall anywhere, and if he's gone home." The wife was so scared of all the kids that it was a thrash. I-If I had kids-- when I was a kid, my dad would come and say, "We're having a robot," I think the whole neighborhood would be going, so I stood there, said, "Oh, the robot will be! We'll have a robot." And I have kids, so they came to visit Arizona and everyone there, they say, "Pa wants a robot to buy. It's a horror!" So here. I showed them that post-code, by the way. Remember who, where is the game, where he is--
A man?
Yeah, where he's a man, where he comes with a knife. I say, "This is not a real role. You can teach a robot, there he is, that's, I say, it's gonna be, which means. But, uh, my wife was in a very negative mood, and the kids started copying it, that story is negative. And when I told my wife, by the way, "Look, it's a good story, we can teach him to do one thing. I don't know, wash the toilet. He'll be sitting with a baby-- every day washing toilets all day." "We've got it all cleaned up," that means. She told me that. I said, "I'm not laughing," so she was told that, well, the point is not how it was cleaned.
The point is, he's undisputed, and he's not gonna be screaming or, uh, not telling me, "We're done." Here. And, uh, that's why I'm still a robot, by the way, waiting. Someone wrote to buy another robot. I want to say at once that I'm generally waiting for Optimus. And there are robots that cost strange money. Now, there was a robot showing up, and he was worth $100,30,000 at the exhibition. He has such a perfect woman's face. You didn't see, did you? He's being pressed, he's just like a real woman's face. But I thought that robot could never be home. And this-- he's, by the way, fired as a robot they--
Half won't like it. Pauline's not gonna like that robot.
So he's a nice face, huh? He's a nice appearance. You're...
Come on, nice man face. What's sexism?
But he looks, but he looks, he looks, he looks, this robot, he looks really scary, to be honest. Because when you see the perfect woman's face and you're as low as you can see this metal carcass, next to the neck and so on, the too I'll be honest with you, it's causing this... I mean, it's a thing to see when you see that robot that Optimus or a robot that's the 1XTECH that Neo-- I remember the name of it finally-- that Neo, they-- you know it's a robot. Well, he looks like a robot. And here it is like a real person. And it's really scary because you don't understand the pattern of behavior. And I don't think there's a very good movie on this subject in terms of what happens to these robots.
I keep thinking about this movie when you talk about robot. That's where she tried to drown a baby in the bathroom, right? That's the perfect woman face--
Wait, I don't remember. Maybe he's not. Here.
And this thing is called the effects of a vicious valley.
She fell in love with her owner and decided to get revenge on her wife.
And drowned-- tried to drown, closed in the bathroom and tried to drown their newborn child. After that, I don't want any robots in the house either.
Ildar, what was the effect you were saying?
And what you described, what Optimus, like, looks like a robot, you know, it's a robot, and you're okay with it. And this robot with a human face is too human, and it causes you an internal fear. This thing is called the valley's evil effect. It just happens when the object is too human, but in some detail, you see it's different. You may not even notice these details, but subconsciously you realize that something's wrong with him. I understand that scientific justification here is like a tribe, everybody's acting like a plus-minus, you get used to everything, but someone's sick and acting differently. Not like, uh, you're used to it. And you're starting to notice these things, aren't you? That he's always been acting in the same way, and it's starting to be a little different. And it's causing you inside, well, fear, causing you inside, there, some attention to the height or something. And that's called the effects of the evil valley. He's very often in computer games because when he's trying to do, there's a character that looks like a man, and he's still different. The lips are normal, and the eyes are not lip-related. When a person smiles, like, you know, when a character smiles, the lips smile perfectly, but the eyes are different. And it causes the opposite, this feeling of such fear, something like that happens, something else. And that's called the effects of the evil valley.
Anyway, when I was a kid, I had pulp dolls that were too much like real kids. And it's just me, I'm having a childhood injury like some clown or something, and here's for me these dolls. And I think it's the same thing with robots.
Thank you very much, everybody. I'll see you in a week.