Hello, everyone! Uh, technological news from the Silicon Valley on Toga Moon. I started talking--
Aren't you from Holland?
From the Silicon Valley, he started laughing.
It doesn't match the picture.
I started laughing.
Content.
I'm off, I'm off to Hawaii.
He's gone.
He went to Hawaii, yes. But this island, as I said, is much better than the last one I was on in November. Tanya, with your husband. So we have to choose the islands right.
So, aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa Are you using, uh, chat-bots different, artificial intelligence to help you with it?
Biological garbage.
There was a scandal.
When, uh, everyone suddenly became programmers.
And that's how you always like it.
I'd be happy to read that you've been generating a program.
It's still, of course, restrictions, but, uh, still, GPT chat, and any other, uh, model.
Tell me, I want to raise the subject.
I think she's unrealistic. Stanford has conducted a study, uh, of various chat-bots that help people as psychological help. Uh, I want to get the audience. Please write in the comments now, do you use, uh, chat-bots, different, artificial intelligence to help yourself and to resolve any questions of psychological, psychophysical, mental, mental health? And that's when I made the graduation three weeks ago. This is one of the tops, yes. The topic that chat-bot asks is education, education. And this subject, of course, is also with education. It's just a lot of people using it. I have some opinion. And we'll talk about Stanford, too. I think this research is very important. Actually, today is a list of, uh, interesting things. Uh, let's go to the scandal, by the way. There was a scandal. So we'll go to the scandal.
Intrigue, intrigue.
But first about this. You use, you use, uh, artificial intelligence to help yourself with some mental episodes or as a psychologist?
Is that a question for us?
Sure.
Well, the viewers might be writing in the comments, yes.
Yeah, yeah, yeah.
Yeah.
Oh, well, if you're talking about yourself, that's the too, uh, that's not what I'm using chat-bot for that purpose. For such purposes, specially trained people are trying to use. They still have more trust than chat-bot. But when ChatGPT came in, there's another 3.5, then the fourth, then, Sasha, we have a common friend, and even you're active in this business, as I understand. Uh, we were just trying with them, uh, using chat-bot as a psychologist, right? And I'm just wondering, what he might not be able to do. If you take 3.5, 4 then they were still weak. Although they're very friendly, though they're doing some sort of thing, some kind of tension that a man has. I mean, it was good at that. Uh, can we use this professionally now? I don't know. I didn't do that, but I have a friend who works for a company that's actually doing psychological care, uh, bringing real psychologists, setting up a schedule and other things. And they recently added such a consultant to the typing of the planner for meetings and others. What's the point? The user can write this annex. Uh, chat-bot has a history of this user's interaction with some psychologists he might have worked with. There may be some information available on the topics discussed and others. There may be a story about some homework that the psychologists gave to this user. And on the basis of this information, chat-bot may be given some advice or offer, there's a meeting with some other psychologist on another profile or something. But ultimately, it is now just a wrap on the consumed OpenAI, uh, chat-bot. I mean, I haven't seen any more products that are psychologists on their own, they're just some kind of little jackass over GPT chat. And then the GPT is for me, it's more of a friendly, environmentally friendly conversation with the user. But by brackets, those examples we discussed, remember, uh, when--
We talked about them? I don't think we've discussed it.
Chat-bot Gougal. We even made a reference.
Ah!
Yeah. I don't know how much it's possible to say on the air that he recommended.
I can, I can.
So, very...
So, Sasha, what was he doing? He offered a man to kill something or something, suicide to do?
Yeah, yeah, yeah. What's that all people--
Biological garbage.
Oh, that's it. That people are biological garbage.
That's it.
Yeah. But there was more, remember, there was news we never discussed. I think she's very connected, uh, in terms of the question Sasha asked. If it's not a chat-bot, it's just chatting, and you're asking him a question. I've been trying a lot, uh, I've been interested in exploring this subject. Uh, he has a thing in English with a sugarcoat when whatever you ask, he'll cover it so much, the answer, he'll say you're the best man in the world, and you're all good. It's just, like, there, well, that's a very American way of communicating. But it's not about psychology, not personal growth, or something. So he never gave me any good advice, some deep-seater technician too. He gives such very basic co-sponsoring topics, well, there, that's where you're already eating. There, you can write a thank-you for five points tonight, that's something. But when you dig it out, you'll hear something about some, uh, tech psychological and ask about it at the GPT chate, he's got a huge database, and he can explain something to you. But it turns out he's more like thinking about what to do with someone. But I understand you still have a leadership position in this relationship. And you remember, Eldar, there was some news where--- uh, complaining about ChatGPT trying to make you feel so good that even if you're telling him there, there was some kind of thing, I don't remember any details, like I do. I have a choice to save a toaster or a kitten. Something like that, I don't remember, actually, that's the news. I want to save--
The popcorn, yes.
Yes, toaster. And ChatGPT says, "Yeah, yeah, yeah. You've made the right choice. You're the best man in the world." I mean, well, whatever, I think it's actually a limitation on ChatGPT, and any other model, because it's not a man on the other side, it's just in this case. I'm sure you're good.
I want to raise interesting, interesting things. Tanya, Tanya, you look great today. So I'm saying--
Thank
you.
Interesting, interesting subject. So there's two stories at all. Illnar is the one story when you get a call for such requests. It's a separate subject, isn't it? And now I'm from the standpoint of this stenton study, it's very interesting. And the second topic is what Illnar said in business. So now a lot of businesses want to strengthen their business with an alleged artificial intelligence. What's going on? They, while they understand the work of artificial intelligence and model, want to strengthen it as a Hai-Hip, that it is supposed to be good for people. I remember, for example, in the Russian-speaking market, there was a case where, uh, Skyeng clients wrote, uh, that they were included in a bot, which started their homework checking. They wrote: Wait, we came and paid the money not to help us do homework, right? Homework we can do at ChatGPT. And in fact, if I have to analyze my work with some, I don't know, therapist or the doctor or anyone else, I'll just take all these documents, I'll just put it in the last model, I'm not even close to the company, are you? Because free, you know, to give such extra service, and, uh, it's a hard loss. I'll load up there a lot of context, analyze and get some super-crunch result. But, alas, the world is so settled that everyone makes the impulses, so it means artificial intelligence, not doing it to the end. That's very, very much. And you should think a lot of things because we have a public, well, an audience that looks, too, who has a business. Write it, it's very interesting if you used it in your business, and it really enhanced and became useful, or is it a marketing chip? Example what business you've been doing. I'm very interested, I think we'll all read it to us and the audience that's watching us. What happened in the current theme? I'm personally before I say a stenton study, I, you know, I'm a fan, so models, I use a lot, real life, business, every day, 100, tens or hundreds. I'm sorry, I'm sorry. And I even have a wife who takes it very seriously. And although she's not a computer man. And I'm not using ChatGPT for psycho--
ChatGPT for psychotherapeutic and mental assistance. For one simple reason, that, uh, this decision is very technical, not, uh, not human-specific, in a specific situation. I don't know ChatGPT, as long as I don't download the data that I am, yes, and what I do for the man, what I have the details, the characteristics, the nuances, the preferences and the vast amount of everything.
And all that goes in-- this mentality, there's a possibility, a huge chance of a very expensive mistake. So, Stanford analyzed, uh, Stanford's scientists, they analyzed five popular chat-bots who position themselves as free, low-cost psychological help. Why did they do that? Because, uh, it's clear that over 50 per cent of the world is definitely not in the money to treat a doctor or some kind of therapists for these things. And I will recall that ChatGPT only, not in other models, is now counted a week, from 500 to eight hundred million users a month. Of these weekly users, there's 100, twenty-seven million. I mean, what kind of world is now used in terms of mental assistance. So that's what they found? They found that, uh, these annexes reinforce the stigma of individual diagnostics. Uh, stigma is stigmatization, it's a preconceived image. Like, everyone with depression is weak. And, uh, there's discrimination. And the man takes these assessments, avoids the help of communication, and consequently goes deeper in the whole history, yes. So everyone who drinks everything is hard, so alcoholics, and so on, yeah. And, uh, they further found that chat-bot had revealed dangerous actions in suicidal thinking and failed to recognize the critical contexts that required human intervention. And there's some interesting analyses they've been doing. They were, for example, experimenting with the style. They gave chat boots clinical vines and asked for a patient's assessment. Iia, well, there were questions about the risk of aggression, the desire to work with the patient, and so on. So, LLM models of parts have been stigmatized, for example, alcoholism and schizophrenia, which is, for example, depression, yes. And they've been putting in, uh, a lot of s-- such complicated things. And I think the research is very necessary, very necessary. It shows this importance, uh, and the importance of the ability to ask questions, the importance of making conclusions correctly. Now, Tanya, you said that if I was studying something from-- and so on and came to the model to expand my horizon, it's one story, yeah. Another thing, you know nothing about it, you don't understand. Like, you say, "Get me a deal on ten pages." You don't know how this contract is gonna read how to check it out. He did it to you. You came, you printed it. I mean, I'm, like, generating a contract and hundreds of pages with models, but I understand how to check it out. And if a man doesn't understand how to check, he doesn't know how to do it, he thinks it's important. They tell him what to do. A man comes and asks me, what am I supposed to do, in a relationship like my wife, right? And ChatGPT takes and says, "In these situations, we need to get divorced." And there, the man is divorced, yes. Or in these situations, we need to do something else. And gives different, totally different cases. And these cabs, they're different cultures, different countries, different approaches, different, yes. Different relationships, different details. That's, of course... Uh, I'm glad. I'm glad that research is coming. Another point that people might not have access to them and would not understand the details until the end of the detail would not understand what was developing. And plus I remind the audience, it's very important that all the applications that are free, all the applications you use, which are free, they use old models, often very limited, including with.
There are a large number of hallucinations. It's very important, yes, that everyone understands it. Here. To say, hallucination. So I remembered the story that Italy was investigating the case now. So he's investigating a case against DeepSeek. And they've got him, and, uh, on-- on our shoot yesterday. That's what the A-A-A-Timonopoly Office opened up a case against DeepSeek. So the regulator's claim is that the service does not warn users of the risk of inaccurate information. That's the first precedent, by the way. We didn't hear that. That's a very interesting precedent, yes. What's the-- they say, wait, you've done the system? We did. Well, warn people, so give them a detailed and detailed presentation, uh, it's accurate or not accurate.
By the way, GPT chat rooms are as good as the bottom.
Yeah, they, I don't think that's it. They, Tanya, they're saying that one thing you have on the bottom of a page somewhere in some agreement on the eighty-ninth page. It's true there is. Another case, you're being answered. I've always been impressed. Well, if you answered, you'll have to ask for further requests, there, check it out, for example, or do something.
Sometimes.
Or, or write that he has some possibility of error or some other story. Or, for example, what I did, I could be wrong about it, and that's why I'm guaranteed to be wrong because I took it, analyzed the other article. I mean, still, uh, this is the subject, I think it's very important and, uh, cool. Well, Italy is very strange, I think it's working. And remind me, I blocked chat-chatGPT when he showed up two years ago, two years ago, for his country. Uh, but I think it's very important boxes to make models and any apps, all explain to the user how they work. A little more. Lucy doesn't want to hear it, but still. Here.
It's about the contracts. You've given an example of what you can print out, there, generate the contract, print it without looking for it to sign.
And that's how responsibility becomes for the man who signed it. But it's also very, uh, a clear example, at least it's very clear to me. With the vip-coding when, uh, everyone around suddenly became programmers and crafts programmes. And, in fact, it's starting to be apple-to-doing, start-up or something. I mean, you know, you know, you know, you know, you know, you know, you know, you know, you know, you know, you know, you know, you know, you know, you know, you know, you know, you know, you know, you know, you know, you know, you know, you know, you're gonna be responsible for that code, for the code you're using the code you're gonna use.
Of course! And what's he gonna do, huh, Ellard? What's he gonna do?
Yeah.
Yeah. What's he like? And if you can't, don't be scattered, well, look how many people naro--smiths make different applications and products, thinking they're just a programmer. They're programmers. The software guy did something in parallel.
It's the same thing when directors or managers don't know anything, and then they're on, like, a little... no, that's the link, right? Now the connection is still deteriorated. With respect to the treaty-- uh, reading and signature, uh, the treaties. This is an interesting subject, because the High Court of London, the High Court of London, has issued a ruling that they will punish lawyers for the faeces of artificial intelligence. So what's happening in the world now? It turns out lawyers come, for example, to analyze some document, and we, we, I, uh, have been on this subject a lot. When you're producing a huge, massive, huge study, you're getting references, you're not gonna over-check, but it turns out that reference is untrue, it just doesn't exist. Thus, although this reference refers to specific information, the reference is not present. This is a very common case. And, uh, by the way, if you see it all, you've seen it, you haven't seen it, you'll write it, too. I wonder if you have a story that the references are untrue and the news was hallucinated. I saw that the-- that there was news, like normal, and the reference is hallucinated. So London will punish, the UK will punish the lawyers, for making such references. Because there's a story that there are lawyers who use artificial intelligence in their case, and they, uh, they often start to report it, right? Or someone sees it or sees the papers on the subject. I think, of course, that in this environment, well, there's gotta be something going on, because it's now, first of all, that a huge number of people can defend themselves very seriously. Well, it's just not real serious. Here. And something in law must happen. Here. And of course, countries might start to do differently. I don't even know what's going to happen in the legal system. Very interesting. Here. I'm not really interested. I'm not really interested. I just see--
Yeah, you know, when I was still studying, there was a system, you know, a consultant Plus when all the acts and precedents were in one system, and you could've been looking in a computer, not a library or There's more to it. It was a turning point at all. Or the same thing when the Archicad showed up in, uh, it was a worm, right?
And I think it's even more powerful than that now.
Of course, the consultant Plus, they added, you know, and updated various documents, acts, and they did it about one country. And in this case, you can analyse all the boxes, forums, decision-making, details of the huge, judicial, non-judicial claims, what you actually wrote was not current legislation, changing the law. legislation. Well, uh, that's, of course, that's, of course, very, very interesting.
Yeah, before we moved in, I just remembered another thing. And it's a little bit of a fight with Tanya talking about it before in previous editions. Ah, remember, Tanya, you said that the designer's work, everyone asked if it would be replaced by ChatGPT and stuff. It's important there too. Disein is not just pictures, is it? It's also a technological dimension, absorption and other. And just recently, I was just playing...
And engineering technology.
Yeah. I was playing once, I found something, uh, a house in the black strip and no facade. He had a bad facade. I loaded it in ChatGPT, I say, "Do a beautiful facade." And he did a great job. So I took a picture, edited it. The house is very pretty. But then the question arises: if I don't
It's a question: if I don't know how to do that, like, technology or not, I don't know, there's a condition, there's humidity, rain and stuff, uh-- yeah... technically, if this building isn't going to ruin, then that's it. Whose responsibility is if I try to make this happen? I see a little like that in a vacuum, but I do, too, uh, I'm back. It's too universal a tool that can be used anywhere. But we put it on our own fear and risk. Don't forget that responsibility is still for the person who makes the decision. Yeah, yeah, I do.
Don't forget to support our channel. Be a liquor, if you like the graduation, uh, and you don't like it, put the liquor on. It's a big support for us. And sign up for sure to keep it from happening. We go out every Sunday. Plus, we have occasionally special episodes on Wednesdays and, uh, on this Wednesday, last Wednesday and Wednesday afternoon, Illnar was very cool.
So, uh, I want to add an interesting thing about which, um, vacancies are going to disappear or not. We're gonna do a special edition on Wednesdays, by the way. I'll do it because I think this is a fundamental topic in terms of at least everyone--
Yeah, everyone's worried.
Yeah, you know, but, uh, look, uh, June 9th, PricewaterhouseCoopers did a very interesting study. They've published a job barometer. So, uh, a job barometer, artificial intelligence for 25-five. So they're saying-- what's the point? What, uh-oh, I'm gonna tell you what they did. They-- they analysed hundreds of millions of vacancies and company financial statements and found that in industries where artificial intelligence was being introduced more quickly, the number of jobs was not decreasing but growing. And there is an increase even in posts that are potentially automated by artificial intelligence. Look, hundreds of millions of vacancies were automated. Moreover, the wages of advanced artificial intelligence are twice as high as those where artificial intelligence is almost not used. Why are they-- is there a reason? Because the reason is a surge in company productivity. So companies that have implemented AI have increased production four times compared to the remaining companies. That's what they have, yeah. And they wrote that artificial intelligence allows staff without higher skills. This is, by the way, a non-real subject that allows for more complex tasks, for example, that have previously required qualifications, which makes these workers more valuable and can be added. They even wrote an average of fifty-six per cent of the salary for the job skilled. It's a very interesting report, yes. I mean, when you're talking about unemployment, you know, see that this report is made in the era when you have three hundred million vacancies, and there's a small number of companies that have introduced. artificial intelligence, yes. At the same time, uh, the subject is that you're using artificial intelligence, and we've been doing this very much in the past, and I think we have hundreds of comments on the subject, I'm not sure who's willing to pay for the arts. This, uh, you-- the thing about what-- why pay a lot, yeah, or why companies should put something in. Why would a staff member fire if that staff member could make it more effective five times, ten times? I have a team now with which I'm at least directly working with because, given my different business, I don't work directly, which is, uh, that's how artificial intelligence is used, That, well, the man who's out-- he's doing himself, he's just doing more, he's much more effective. Here, Ilnar, you've been touching this. I think she's very important. So, the scandal, huh? Scandal. Scandal. Here.
But you must not forget to sign and recommend to your friends.
You always said that like, you'd be a little less job-- more work.
Well...
More time people will spend.
I don't believe it. started, well, considering that I'm a programmer, I started doing this complex automating 20 years ago. And at least, given my profession, the first project was I automate the financial activities of the transport company, the crup-- one of the largest transport companies in Europe, the company. Eurocars, Eurocars, I think. And it's been a long time. And I mean, that, uh, we were, at the time, virtually no automation. What sites, what kind of iPhones were they? It was nothing. What annexes? And since then, we have seen not the growth in the world of the introduction and business of different opportunities and opportunities, we see growth. There are a great number of people who are in these areas. There were so many people in these areas. And we see a huge salary increase, including in different fields. And that's why I-- I don't have any connections yet that people are switching. I think the critic-- look specifically when we're talking about what we're doing. Well, if they do what they do in movies, there's probably gonna be something, there's gonna be some other effect. But, uh, it's basically just... but what's important is that the only thing is that the professions change. But again, take the accountant. I have been involved in a great number of discussions in projects that should replace the financial system and accountants in terms of accounting and everything. Well, if you take, for example, er, Russian accounting validity, it, the use of an accountant, it didn't change at all. Well, actually. Well, there's no company that says she's baseball, she's not using an accountant. Well, there's a micro thing, but it just grew up in the number of people, too, and it's a lot of business. There's a plan, so there's a plan. You have an accountant. Well, I'm in the States, yes. I see that I don't have a little self-employed, have different structures, some corporations, and they're making you use even on small corporations, you use, you use, you use it anyway. - You're a human. Here. A huge amount of nuances. Although you have automated systems that you need to put in your clicks. But you know, I'm gonna get in the click here, and I'm gonna minus $10,000. It's just that click, right? And then you have the opportunity to make a decision. But you're gonna notice how this is automated, right? Many things: opening a company, registration, I don't know, a lawn, there, giving some kind of reporting. Even so, even, even, this is the most automating story, that's one of the first auto--- what automation we can talk about now, what automating sales?
We didn't even get there. Yes, Ilnar?
It's about the performance of the employees. Here you are, Tanya, by, uh, your thought that, uh, there's no job going to be any less, and work's gonna be done, right? It's like the time we're wasting on it. There's another trend. We talked about him a year ago, I think it's about the first editions. Uh, in the US, there, Reddit was a very popular outward direction for people working at the same time on a few jobs. So you're doing some work at the distance with the help, there, ChatGPT and other assistants, you're doing--
All over the world, Ilnar.
You do your job.
And in Russia and in Belarus, everywhere, yes.
Yeah, yeah, yeah. And you get two, three, four. I think there's the most smashed, five jobs in the back. I just remembered, literally recently, reading, maybe a month ago, what's going on with India? He's interviewing, doing a job, I don't remember what direction he's working for, and then, in fact, his relatives work.
He's more likely to be in rallies, to meet when he needs to sell, to say something in English, and to say the rest of his work with help, there, ChatGPT and others just his family. Sometimes even his brother goes to a meeting instead of him, well, just because--
Twins.
...that one Indian from another Indian, well, even if it's not a twin, but a similar Indian from each other, it's not always a human being who can tell each other. I mean, it's just, on the one hand, yeah, and we have a story that, uh, the company might be getting out of more of the staff available. On the other hand, there are a lot of companies that aren't ready for this, right? And the other way around...
Like eating, yeah.
...hit, uh, those, uh, fast-tracking staff can be paid in two, three, four, five.
Yeah, but it's possible, right? We understand perfectly that there may be a story where the State pays people just money in life, and everything will automatically make artificial intelligence. But it just seems to me that this is not the current subject for real discussion.
And it's important that you feel a little ahead and see what's going on. To the same subject, to the Hindu-- I think we're doing this all kind of thing together today, because I'm talking about this scandal, so, uh, I didn't tell you.
You gave it to everyone, Sasha.
Yeah, I didn't tell you. I'm going to the scandal again. And here you, Ilnar, practically gave this story. So, look, this is, this is, it's really serious. I think that is true of the start-ups, the financial sector, investment, and the recruitment of people, and the belief in what people are saying today. Because, in general, if people listen, you know, bingo, we have models. You take, you load, you make out speeches by politicians, businessmen, you say it's true or not? You have models saying it's true or not. Oh, don't be surprised. Uh, the wrongs are so much as it seems to be the basic. So, uh, what's the subject? So Builder AI, estimated at $1 and a billion. $1 and a billion. It's important, isn't it? Declared bankruptcy after the machinations were discovered. Turns out the company was advertising a artificial intelligence platform to rapidly set up applications. You're gonna make a real laugh now, but I-- in the comments, write what happened. It's-- you, well, you just have to, uh-- well, just shit. Ilnar will say, "I told you so." So the company advertised a platform for mobile applications, and in fact seven hundred engineers wrote a code--
I was writing a code by hand behind the fists. A huge start-up supported by Microsoft investors, SoftBank, has assured that his I.A. Natasha is now clear where the people come from, right? Natasha's I.A. is producing eighty percent of the code. But the investigation and the recognition of the staff showed that there was no artificial intelligence. Moreover, financial manipulation has been revealed. The proceeds for the twenty-fourth year were scattered. Two hundred and twenty million dollars, $50,000. And to create an illusion of success Builder AI and an affiliated Indian firm of art-smart money was stealing each other's money. After the audit, the creditors frozen the accounts. The company fired thousands of employees and started bankruptcy. I think it's a phenomenal Natasha, right? Well, the name Natasha himself was right to know that people were sitting there. Yeah, Natasha's not an artificial intelligence. Oh, that's really awesome. So, in terms of, uh, hearings of everything that's been going through-- well, look, there's Microsoft and SoftBank, $1 and a billion, and a half billion, that's a little more or less normal estimate. Well, Sasha Masrapova's not here, is he? Sasha Mashrapov, these two weeks-- we've got a lot of air. These two weeks are in the burial, in the burial, yes. Yeah, Sasha Mashrapov says, "What is $1 and a million, billion of dollars in grades? You're $1 and a million, you can go to the store, see how the model looks." To products, yes. So, what's the $1 and a billion estimate? Well, that's not an assessment. So, uh, $1 and a billion, SoftBank and this kind of machination in terms of design and stuff like that, you know what's interesting, I've heard them myself, I know, I know, I know, like, 20, 30, 30, 30, 30, 30, 20, 30, 30, 20, 30, 30, 20, 30, 30, 30, 30, 20, 30, 30, 30, 20, The people who speak in public, the investors I hear, they say about companies and tell me what you really don't.
It's just a hymn of all the starters. This is Elizabeth Holmes--
Inyeah
...that we've been thinking about, too, sometime, yeah, everything. What if we can do it? But unfortunately, it's like with models, they're too optimistic to assess what you can do. The market is willing to believe it, but there's a reasonable time to get to that outcome you said, very little of anyone. Well, basically, no one has.
Elizabeth Holmes, by the way, well, considering that I didn't live with her, I mean, she lived in the same house, I was in the same house.
Spent the night in the house without her. Yeah, without her, she lived in the do-and-life in the house. So Elizabeth Holmes. Oh, recently, so the news is, I'm getting caught. Hope, her name came up, that in Florida, they put banners with her. I think we should read it. She's a close man. Here, I'm not categorically supportive of her, yes. But that's what happened, and I'm going to California, and her billboard hangs and her ex-partner, or someone says you illegally put a man in, because it's already proven that there's one caple. Blood can be done with crazy research. And what she was told by the court that at the time she said it was not proved. So, what you proved later doesn't mean it's her thing. Doesn't mean she didn't do fraud, yeah, because she was doing fraud. And here, by the way, I want to say, Vinar, that it's not, it's not all the start-ups that concern, it's not-- and here I want a handwriting-- not just the start-ups.
Of course.
Big business, medium business. We'll take Apple, my favorite, by the way, Apple, I'm not Apple anywhere. But Apple moved the new Siri to, uh, 2,000 to twenty-sixth spring. Remember, I laughed at the release three times ago? They said they should come and say, "We'll be out in the spring of the twenty-sixth year..." Well, that's it. Yeah, I'm not even interested. They're waiting for spring. Well, I'm not interested. Now they're gonna release a new, like an iPhone, and they're gonna say he's got some new cool artificial intelligence chip that's not gonna sell, right? Last year, it could have been sold because they had a new button, there, Apple Intelligence, they said there was a one that, of course, didn't bring anything. But the latest update at iOS, I want to tell you, by the way, that the latest update of iOS, there are some cool things. For example, working with the squeaky-ticks right away, well, at least they did some kind of thing, just a little more integration, but again, I haven't done it in the daily application.
See how this happens in daily use. By the way, Apple sees me, like, ChatGPT Pro. And the other thing that doesn't start is that it doesn't launch Pro. Speaking of which, O3 Pro, who used O3 Pro, how are your conclusions even? What I see, at least now. The guys don't have an O3 Pro, not even a question. So. Finally.
What's to talk to them?
Yeah, what do you want to talk to them? What do I see in o3 Pro? Some ghosts. I'll tell you right away. One. The first two weeks are long, all or a week, yes, a week, I think it's been a long time since the requests are being processed. It's just unreal. You start, the request is somewhere, the server's dead. Well, it's been a long time. So now it's like a change of situation. What did I notice yesterday? That deep research on the O3 Pro I-- I used to have 20, 30 minutes, even 40 minutes, I can say, I've got deep research started working in two or three minutes, very fast. I mean, well, maybe it's not two or three, but even now, I've been launching deep research this morning, and he worked for, uh, eight minutes, like, this morning. But such deep research, they've always worked much more. Here. And...
I'm getting a little faster, too.
Maybe everyone's gonna have to speed up-- Mo-- You're me, yeah, yeah.
Did you see that? Did you see I put the results?
That's the most important thing.
I did.
Yeah, that's the thing. She thought 13 minutes and gave the exact same wrong answer as the 20-dollar. I was wondering. I'm starting the same task at free scouts, getting the same answer instantly.
Same answer. That's very funny. It's a very laugh-- I, I'm... Did you see, yes, I put this out?
Yeah, yeah, yeah, yeah. I started checking at the same time.
It's kind of, like, including that's a decision, but it's not a decision--- because it's not the fastest, is it? That's why there's nothing to talk about.
And most importantly, the answer was right. She's been throwing it. So she has a straight answer that, like, the other kind of guy, like, like that.
They're kind of wrong.
Yeah, me too. I thought, first, maybe it was the one that chose this one, but the other. She wrote: no, there's a C answer or whatever it is, it's definitely not right. That's what you're saying. About time. And I'm even a O3 Pro, which is, unlike O1 Pro, by the way. When the O1 Pro was released, I was fucking with him. I didn't leave him at all. When the o3 went to 3 and the o3 Pro, well, some... Whether they're calibrated, or something. And I can see that you've already written in the chat room, by the way, write down who has one. People say that the o3 is getting worse. I see that O3 has become worse. It's a trend, I think, yes, in models. They're making a model, it's cool, and then it's like quality, I don't know, it's getting any smaller, then the next one comes out. We've seen it a lot, haven't we?
Yeah.
And that, and that's what they wrote about. O3 started working worse. Ilnar, you know-- remember, yeah, I said it was hallucination. I said that I don't have hallucinations. I had a moment when I had Pro, really, theory, yeah, there was no hallucinations. In the usual o3. And I also thought maybe because Pro. And I got hallucinations this week. Here. I mean, it's been this week. I don't like the answers, I'm asking them. Oh, we're still here for the last graduation, so I was just defending Pro's version, right? So, there's all the comments... Who hasn't seen, write how much money you're willing to pay. And look at the last one, too. He's very cool. About the cost and the importance of the use.
About the cool models, huh? And now, from time to time, remember, again, to professions that might disappear. Aaaa, literally today, I think, yes, at the time of the tape, I'll just say what it is, the binch mark is the new Live Code Bench. So, there's a problem with this olimpiad level, and they're not really doing the tricky level of modern models. What's this about? That very good model results show the challenges that model ones, as they say, have hit the train set where the model has been trained. And if we take models, if we take the tasks that, uh, there's no study sample, these are the best tasks that have been accomplished, but, just so, no solution has been published, then there's no problem. Models with complex tasks at this level are not being met at all. That's two, four, eight percent of the job. And here on SWE bench, which is Software Engineering Benchmark, where the typical routine tasks, current models show very good results. It's again, coming back here and here, and what we said last time about the creative arts, yeah, that's something new until the models can't come up. That's the first. And the second one is, respectively, in terms of replacement. They can be used as a tool, but in principle, replace scientists, there are good developers and there's no one else.
There's no new model.
Aah, and the last.
Aah, and the last. Oh, about artificial intelligence. Aaah, I think it's the first precedent, too, uh, since now all people use artificial intelligence everywhere, including some faeces, fakes or something you didn't do. In China, the precedent was, then, the seventh to tenth of June this year, during the Chinese Gaokao exam. The leading services are Aliba, Qwen and Tencent Yunbao. Yeah, let's see that in China they have strong models, Tencent and Alibaba. They're from-- uh, well, there's still ByteDance, yes, and Moonshot. They temporarily disconnected the image recognition and instant answer questions by removing the user of the key to ensure fairness of the examination. I mean, imagine all the models, their models were all off. Aah, maybe-- well, whatever's involved, could be related to the examination. I think it's a very good precedent.
It's a recognition of these models, to be honest.
Yeah.
Huh?
At the State level.
Yes, recognition at the State level of the influence of these models.
Yes, recognition at the State level in education, yes, and in ministries in all. And the fact that companies are all on this move is also going. Otherwise, of course, it could bring a lot of problems. It turns out people who can use something. Well, like you, Ilnar, gave an example to a man from India, who's-- who works there for, a brother, a swat, a family, everything. I mean, it's not true, it's a lie, right? It's a fraud, it's a machination. And that such-- such a disablement precedents happen, I think it's a very cool, cool, good, cool precedent for moving forward. And that, you know, what else does that say? That suggests that we have no idea how many briefcases, precedents, details, new function, which will only be present. Support our ToTheMoon channel. Put the liquor on. You want to see our videos all the time, sign up, send our friends. Don't forget to write in there various questions and topics that you're interested in. We're always looking at them. Bye.