Guerre des IA : l'intelligence artificielle va-t-elle mal finir ?

Good evening, my friends. Two pieces of news this week that show that the battle of Lia, which has only just begun, the battle of artificial intelligence, is taking an increasingly brutal turn. There is a lot of money and a lot of interest at stake. All is fair. And in fact, we are heading more and more towards battles that could end badly for artificial intelligence, its users, the people who depend on it, and also perhaps even sound the death knell for this technology. Because we must know one, there are not only technical solutions. You know, when we look at solutions, it is not because it is technically possible that we can do it. There is also the legal aspect, there is the aspect of compliance with a certain number of rules. I want to say civilizational. That is where we are today. We are at this crossroads. I will tell you about it without further ado. We have Perplexity, an artificial intelligence that was created around 2022. Well, it is the subject of a complaint from the Financial Times, the financial newspaper that everyone knows, and also from another Chinese newspaper. So what does the complaint say? It says that Perplexity, in order to train its model, it copied it massively from the Financial Times. That means it took articles from this newspaper which are original articles which required humans, well, not just to copy each other, but rather to go out and do investigations, ask questions, take photos, do field reporting. Right, write articles. And then what does Perplex City do? It records these articles and then spits out the figures, spits out the information, the data, including the errors that were intentionally slipped into the newspaper, and then gives them directly to the user. So we are in a data theft situation. Between us, AI is always trained by stolen data. That's the big secret of artificial intelligence today. It's stolen data. Data that is collected from just about anywhere. That means we're going to launch bundles, we're going to do scraping, we're going to download texts, download images, download everything we find from more or less anywhere we can find it. Again, we're going to download it online. And, and in fact, we give this, we give this dirty work to intermediaries who are often small companies located in countries where the jurisdiction, the pasta, they look at this kind of thing and then what do we do? We package all this, we use it to train LLMs. Now, what's the danger? What's the problem? We're going to talk more about data theft, which is what we're going to talk about today. But what's the problem? The problem is this. Look at Google in English, for example. Google in English will probably soon be available in French and other languages. When you do a search today, it suggests that you go into AI mode. So what is AI mode? It's a mode where you can ask questions and it answers you directly. So instead, what is Google's base? A search engine. Normally it's supposed to offer you links to websites that you should visit. What's the rule here? You're going to visit and find your information. Except that here it offers you information directly. But where does this information come from? It comes from websites. What are these websites going to see if this way of consuming information becomes widespread? What are these websites going to see? Well, they're going to see that their number of visitors has plummeted and that no one is coming anymore. So the webmasters of these sites will see that there is no momentum. They will say that maybe there is no interest, when in reality, perhaps an article they wrote that was viewed maybe 50 times in the website calendar may have been used half a million times by consumers who have gone through artificial intelligence. And all of this will create a copyright issue, a question of lost revenue, and then in the end, a question of motivation. Quite simply, ah, we can't expect people to post original content, create original content if it is immediately copied and reused elsewhere by giants of artificial intelligence. So this will also end up in lawsuits, stories and others in court, just quickly. This video is sponsored by Kelou's. It's an organization in France that helps people who want to find a job or change jobs, who want to get stuck somewhere, and then they don't want to confront the market. Well, they want to, but they need help, they need support. Well, Kinnis are very professional in this. I have advised them, I have sent them to many people over the last 12 months and I have received excellent feedback in some cases. It is supported by France Travail, my training account. So I will put the link in the description of the video and in the pinned comment. Go ahead. On my behalf. So I will continue with the data theft. We have another piece of information that is in this continuity. We have an engineer from xai. XAI is Elon Musk's company. It is the one that supplies grok. Grok is known on Twitter, or rather what we call X today. So we have A Chinese engineer, Mr. Lee, I can't remember his full name. Right, so I was saying he joined XAI in 2024. To give you an idea of his background, he studied at Stanford between 2021 and 2024. Three years of studies at Stanford, part of which was during COVID So probably remotely. I'll tell you the rest. The rest is that he joined XAI last year. He received several million in salary and also in shares. And this August, what did he do in 2025, he sold $7 million worth of his shares. Do you realize that today? Huh? There are ways. It means that while AI is destroying jobs, it is also creating others that are extremely well paid. But it's gold for the brave, because it's not about knowing how to make a pretty prom or things like that. It's about people who can go into the heart of artificial intelligence algorithms to interact, to try to improve and all that. And they do scientific research in this direction. This guy, Mr. L has been cited. You look on Google Scholar, he's been cited nearly 18,000 times. So he's still someone who has contributed a lot, but in a very short time, they contribute a lot. And the entry bar is quite high because it's equivalent if you want to do a PhD in mathematics. In fact, people with a PhD in mathematics ended up as a mathematics teacher and you'd see them in the morning waiting for the bus to go home. You know, mathematics has never really fed its man, except that today it feeds him. It feeds him a lot. So I was saying he arrived at 11 last year. This year, August 2025, he sells his shares, which sells for 7 million shares. He resigns. He signs a contract with Chad, GPT, if you like, OpenAI, the creator of ChatGPT. He signs with them and we discover in the process that he stole some code. That means that before leaving he took some code. So we don't really know yet what he took, but he is the subject of a complaint. He stole information from his employer. So what does his employer do? File a complaint against him? Not against him personally. Among other things, they want him to be punished by law and so on, but also an injunction. They want an injunction from the court so that he can't work at OpenAI. So if he can't work at OpenAI, who do you think he's going to go to? Well, I can see him getting on a plane, going to China and working for Chinese artificial intelligence. Right. Whether Oni or Kamali thinks. But you see, it's really starting to get ugly. Yeah, it's thank you my friends and see you in the next video. It.