Back to episode — Episode 3000 CWSA 10/26/25
Context —
t marching around with other people who have the same viewpoint would be a form of group therapy. And I thought to myself, you know what? It is group therapy. And you know what else? I'll bet it works. I'll bet it works. Just feeling validated by marching around with people who agree with you and tell you that you're right on point. Oh yeah, you see the world clearly. You don't have any mental pr…
← Previous segment →n AI will never get you to intelligence basically and that the whole pattern, you language, the large language models, you could train them forever and it would never create intelligence.
So the head of AI at Meta is saying exactly what I've said for two years, that there's no logical way you can get from pattern recognition of words to intelligence. But it might look like people. It might look and act like people because what we would discover is that people are just operating on patterns too and we don't have intelligence. Was I right about any of that? Yeah, I was right about all of it. I was right about all of it.
Now what he says, and you've heard other people like Elon say this, I think that the current models, they look like they're going to be PhD smart, but they never will be. There's no way they can get there. And I've been saying that. I was like, I think this all looks like a fraud. It doesn't look like the thing they're doing with these gigantic data centers would ever under any circumstances could ever succeed. It's just logically impossible. I guess he agrees.
But there's a solution. I don't know if the solution makes any more sense, but the solution is that the models have to look at things happening in the real world and understand real world physics. So the thinking is if your AI could just observe humans or other robots doing things, it could learn sort of the way that humans learn by observing and then trying to imitate it. So that would get the robots and the AI to something like real intelligence. But that's untested. It's definitely the approach that I'm pretty sure this is what Elon's doing now. I think he's training his with real world physics as opposed to large language models. Talking about his robots specifically.
Something else I was going to say about that. I guess as things like the robots' hands get more sensitive, Tesla has good hands on their Optimus, says Elon. The hands would capture a lot of stuff in the real world too, as long as it's doing stuff. All right. So that's coming.
So the robots now are being trained on artificial worlds, which they imagine, and I'm trying to figure out if that matches how humans learn. That was weird. Hold on. Oh, you're kidding.
It's the worst lamp in the world. So I've got a lamp that doesn't have any raised buttons. So when the lamp goes off, you have no idea where to touch it to turn on the light. And if you hit the wrong button, it goes on a 45-minute timer instead of just turning on. Who designed that? Who designed a light that you can't find the switch if the light is off? I mean, seriously, did they test that even one time? Did they test their light in a lit room? Every time they turned it on, it's like, "Oh, look how easy it is to find this button." I swear to God, I have so many products that could not possibly have been tested even once. Even once that could not have been tested. God.
Anyway, so these models are imagining what the world looks like and then learning from their own imagination. So the same way that you would create a video AI, you tell it, oh, imagine a cat as a samurai or something. The AI is doing the same thing. So it's imagining a world and then it's observing its own imagined world and then it's trying to learn the physics from the world it created, which I'm not sure makes sense, but people smarter than me say it does, so I'll accept it.
Apparently Apple, according to DogeDesigner, this is the only place I saw it was one account on X. DogeDesigner says that Apple's in talks with SpaceX to work on Starlink, meaning that your future iPhone could work on Starlink alone as opposed to a phone company. Now we're getting interesting. Does that suggest that SpaceX and Starlink will not make their own phone? Now, apparently they're not talking to Android. Don't know why. We're not talking to them yet, but they're talking about it in terms of dead zones. So if the only thing that Apple is doing is agreeing that Starlink will fire up if you're in a dead zone, but only in a dead zone, then it's not that big a story, but it's kind of cool.
I don't really have any dead zones where I live. Do you guys still have dead zones for cell? When was the last time you were somewhere where your cell didn't work? On a mountain somewhere? I don't do a lot of hiking.
All right. So Trump's got this lawsuit he's had for a while against the Des Moines Register newspaper because one of their pollsters, close to the election, this last election, published what most people would consider a fake poll. So it was one of the most respected pollsters, but this Ann Selzer pollster just before the election dropped a poll that said that Kamala was doing great and none of that was true. So people thought that the poll was intentionally fake to bolster Kamala and that's what Trump is claiming. You know, I guess he has some kind of a state court win that is not a victory on the case, but rather allows him to press it. So that the only thing that's changed is that he's allowed to go forward with his case.
I think he's got a case that it was so obvious that that was a fake poll. I mean, again, I'm not an expert at polling, but I knew it was fake as soon as I saw it. Didn't you? There's a lot of stuff you don't have to be an expert about to know it's fake when you see it. That's sort of my only skill. I don't have technical skill, but I can spot pretty easily. Pretty easily. Not every time.
There's a Center Square Voter Voice poll that found out that almost half of adults between 18 and 29 either have not much confidence or no confidence at all in the midterm elections coming up. So half of people in that young category don't think you can trust the elections. That's just MAGA. Do you think that came from Trump or is it just a generic young people don't trust stuff? Could be both.
Anyway, my question would be why would any young person trust an election? I understand the argument for not trusting it, but what would be the argument for trusting it? Like how inexperienced in the world would you have to be to think that there's something like any big organization that has any kind of power over you that also operates completely ethically. We don't really see that. When there's some big organization has power over you, they're always operating unethically. Always.
How's your financial world? How's your healthcare world? How's your news? How's your taxes? How's your professional gambling and sports? It's everything. You know, if you're between 18 and 29, you think you didn't notice that literally everything is rigged. And to believe that the only thing that's not rigged is the thing that is rigged the most and probably has the most access to rig. Probably. Really? Is that what you believe? That everything else is observably provably rigged from the lawfare to the government. You name it, right? The climate change was a hoax. All the news is fake. Everything's rigged except the elections. How can you believe that? Like how inexperienced in the world would you have to be to believe that everything is rigged provably so except the elections? I don't know.
People, today's the first day of early voting for New York City so I would say that the result is over. There was some Hail Mary hope that maybe Mamdani would drop out before the voting started. He did not. And apparently it was just a huge day of fir
Context —
st voting, just lots and lots of numbers, which would suggest that the conclusion is already a done deal. And it would suggest that Mamdani is going to win easily. We'll have to wait for the result, but that's what it suggests. So I wasn't really following this whole Canada and Ronald Reagan tariff advertising story, but it just persisted a little bit longer than I wanted, so I'll mention it. So…
Next segment → →