ront of you. Now, I don't know if they can do it with photorealistic stuff. They did it with an animation. So you look at the table, and sitting on the table apparently is a little character that's dancing right in front of you as a 3D object. Kind of amazing. I didn't think that was going to come.

Well, I did a little deeper dive on quantum computing because I was trying to understand. I've told you this before. When I hear people explain how it works or why it works, it sounds like word salad. And I thought to myself, hmm, I guess maybe it's because I don't have the technology background, but if I did, I'd understand that word salad. And I went deeper and deeper, and I'm quite sure that nobody knows why it works. I'm fairly positive. And you've heard things like, I think Joe Rogan said this the other day, that quantum computing works because it accesses the infinity of other dimensions. Nothing like that's happening there. There's nothing like accessing other dimensions. That's not happening. We don't know what is happening, but there's no indication it's accessing another dimension, if there is one. What is it even?

So the things they say are things like, well, you know, this probability wave interferes with another one, and between the two of them they can predict the future without using any logic or any calculation. They just see the future. And that future they see could be like the solution to a difficult math problem. And I say to myself, really? You just interfere a couple of probabilities and then you can see the future? So what quantum computing is alleging to do is literally see the future or peer into the past, I guess, too. None of this is really possible as far as I know. So I thought to myself, it must be me, right? It's just because I don't understand the field. And I've now seen two experts who do understand it kind of suggest that nobody understands it, and we don't know if it could ever even be commercial in a big way. I mean, it might have some niche specific things it can do, but it's never going to be a general computer. So forget about having your own quantum computer. It doesn't look like that's going to happen.

But Nvidia just created a personal AI supercomputer, and it's only this little box, which if you wanted to work on AI, if I understand it right, it would be like a data center in a tiny little box. So if you were working with AI, instead of having to access a data center, you can access this little box on your desk, and apparently you have full AI capability. So I don't know the ins and outs and the pluses and minuses of doing that versus having access to the actual full data center, but it's got 200 billion parameters, and it looks like a big deal.

And then they also have a secret chip that's going to go in humanoid robots. So with all those new introductions, I guess the stock for Nvidia would be, and it's down. I don't know, maybe people expected more. Seemed pretty impressive to me. I think Nvidia is worth well over three trillion dollars now, and they're talking about things that look like it would have a big impact on Tesla. Because if they put these humanoid chips in robots, does Tesla buy some of those chips, or do they have their own chips for the robots? So I don't know. Who's smart enough to know if any of these Nvidia things have any impact on Tesla's future? But I don't know. We'll see.

In The Guardian, they're saying that virtual employees could come this year. In other words, an AI agent that can take instructions and act like a person and get things done. I'm going to go contrary. I'm going to say it won't happen. And that's based on my experience where I tried to create, over the summer, I tried a lot. I worked pretty hard on it, trying to create an AI agent that could just answer some questions on my Dilbert.com web page. And what I learned is it doesn't look like it's possible. And here's the problem with the current AI, which would be different from the future versions, which are AGI. But the one that uses a large language model and just uses patterns, it can't answer the questions reliably because it hallucinates, and you can't stop it.

Even if you have a database and you say, "AI, do you see this database?" "Yes, I can read it." "Can you tell me something that's on the page?" "Yes, here it is." You go, "Okay, you can definitely read the data. Here's my instructions. When I ask you a question, only look at this database, which is just a document, and only answer from that." "Okay, got it." And then you ask a question, it'll just make something up without even looking at the document. So that was an uncommercial product. There was actually a launched commercial product that as far as I can tell doesn't work. And I don't think there was any competitor, so there was nobody who had figured out how to make it stop hallucinating. And as far as I know, it might not be possible because the large language models don't just follow rules. They've got a whole different mechanism. I don't think you could make that reliable. I think by its nature it's unreliable, and I don't know that you could tweak it, fix it, patch it. I don't think you can get there from here. So that's why they need the AGI. That would be a whole different technology that we haven't seen yet. I don't think it's invented. But I guess Sam Altman thinks we're getting close and it could happen pretty soon.

But I'm going to go contrarian and say the current AI is not going to create an employee that you can rely on. So I'm contrarian there.

All right, so there were at least two things that were fake news that I reposted yesterday without knowing they were fake news. Now, how much worse is this going to get? How many times am I going to get fooled by something that really looks real? It didn't make much difference. One of them was a parody that