. We're about a year away from robots doing everything that men can do that requires strength. So we don't have the strength thing going for us. But at least we're the only ones who can make sperm. Okay, we're not the only ones who can make sperm now. I guess the scientists can whip it up in the lab. Or whip it up in the lab, if you know what I mean. No, really, it's from stem cells. How do we know it's from stem cells, though? I don't trust those scientists at all, if you know what I mean. Anyway, men are obsolete. We'll be in about a year.

So here's another one of my unbelievably good predictions. So good I wonder why you listen to anybody else. Remember when I told you that I sold my Nvidia stock a while ago because I thought there'd be some major disruption, some technology that showed you didn't have to do these big data centers and whatnot? And sure enough, DeepSeek comes out, the Chinese open-source AI that's cost 5% to make and it's almost as good. But I also predicted, and here was my more clever prediction, that you shouldn't be that afraid of it, at least in terms of disrupting America's AI. Because first of all, American AI would keep moving forward and probably they have some advantages because of the big training facilities that they have that DeepSeek does not. But I also told you that the government of the United States would organize against it to make it illegal or there would be lawsuits. Basically it would just be tied up in all kinds of legal maneuvers like TikTok. You know, TikTok being caught up in the government political process. So I was pretty sure that it would be heavily attacked. And sure enough, here's what we know. OpenAI is looking into complaints that maybe it was trained on, just looking at OpenAI. Now I'm not sure I believe this next part, but it comes from people who know way more than I do. Apparently there's a way to train an AI model where if there's already a model that is trained like OpenAI's ChatGPT, you can have the one you're training just ask the one that already exists millions of questions, and then it just answers the questions and then the new thing is also trained. Does that seem real to you? How in the world could it ask enough questions? I get this as fast, and there's no human involved, so it's digital. But can you really connect to OpenAI if you're just a user and ask it millions of questions a minute until you've sucked all of its knowledge? That doesn't feel like that could happen. Wouldn't OpenAI have guardrails that say you can't have that much traffic all at the same time? Or it would be so expensive nobody could do it. I know there's something about it that's not connecting in terms of probably just my understanding of it. But I don't know how it could possibly ask millions of questions and then reproduce it. That doesn't even seem slightly possible based on what I know about anything. But we'll see. The smart people say that's a thing.

So OpenAI is going to go after them for stealing their IP because OpenAI says we're seeing indications they may have trained on our IP. Which is interesting because a whistleblower recently died under mysterious circumstances for allegedly having evidence that OpenAI may have trained on some copyrighted work. So you've got the one that may have trained on copyrighted work