Anthropic AI Model Hardware Standard for Robots - Your Dishwasher Will Be Conscious

Eli the Computer Guy (0:00)

I have to say, this is pretty darn cool. To be clear, it's not two trillion dollars of really darn cool, but it's pretty darn cool nonetheless. Anthropic pushes into physical world with new standard to help AI agents operate machines. And I think this is really awesome. And now, when I say I think this is really awesome, I know I'm gonna get some pushback because people are gonna be like, Eli, Eli, I thought you said humanoid robots were stupid. I thought you had all these complaints about Tesla and all that kind of stuff. Why are you changing your mind now that Anthropic wants to start operating machines? Because the important thing to understand in the real world where people actually have to, you know, do something, you know, accomplish something, the reality is, is that we do actually use a lot of machines. A lot of those machines can be automated in one form or the other. And having artificial intelligence be able to automate those machines could be incredibly useful. So think about this with robots, right? So in 2026, when we talk about robots, people think about humanoid robots, they think about dog robots, they think about Model Y computers, they think about drones flying into another country and killing people. But the reality is, is the robot that most of us use every single day is the washing machine. That is an amazing robot. But one of the things to think about with a washing machine is that there could be a lot more intelligence used in the washing machine system usage than many of us actually do use, right? So think about this, washing machines many times use a lot of hot water. So one of the problems you can have is if you turn on your washing machine and then you go take a shower, you might get a cold shower, or you might get a lukewarm shower, that type of thing. The problem is, is you go take a shower, you didn't turn on the washing machine, and then after you're done with your shower, you have other things to do, you forget to turn on the washing machine, and then your wife comes home, it's not a very good afternoon. I meant to turn on the washing machine, damn it, right? Imagine if the washing machine had intelligence, right? You got cameras, you got other things going on, it can see that you just came home from a run. So I go out there, I run, I did 18 miles, two days ago, I did my 18 mile long run, I have another marathon coming up in two weeks. So let's say I go out and I do my long run, right? So I've got all these systems that are tracking things. I got my little iPhone, my little iWatch. So my Apple Watch tracks that I just did an 18 mile run. So I come in, right?

I put my little stuff into the washing machine, and then I close the washing machine, and I hit, essentially, the start button. But imagine if there's artificial intelligence that can look at what's going on. And say, well, Eli just finished an 18 mile run, so we know he is going to be taking a shower soon. And after you have an 18 mile run, you don't want a cold fucking shower. That's just not a good thing. So what if you could say, it knows that a shower is probably gonna take me about 20 minutes or whatever else. So when I hit that start button, it actually doesn't start now. It actually is able to pull in a whole bunch of additional information and decide on its own when it should actually trigger, when it should actually start. That could be a really cool thing. Or one of the things I think about, again, being married.

Wife always wants the washing machine, like the stuff in the washing machine cleaned at a certain time. But you never, it's like the whole loading the washing machine. Do you load a washing machine halfway and then run it? Do you wait for the entire thing to be full?

What if you could just tell your little AI system, look, I need the dishes to be done by the morning, right? So when I wake up at six o'clock in the morning, I want the dishwasher to simply have a clean thing of dishes in it. Well, imagine again, every time you come in, you put your dishes in it, you close it up, make sure it's actually sealed. It can have its own little intelligence going on, and then it'll just trigger, it'll just run. Be like, oh, they need to have clean dishes by six o'clock in the morning, whether it's a full thing, whether it's a half full thing, it's going to run on its own, right? I think there's a lot of interesting things that you can do with automating machines. Again, think about things like garage doors. Should garage doors open? Should garage doors close? Access control, I've talked about this for years. I've talked about bringing active directory into the physical world. When you go up to go into a facility, are you allowed to go into a particular room, right? When you go to open the door, does the door open, or is the door locked, right? So imagine if you have an AI system. An AI system recognizes that you're an executive for a company. So even if you don't have specific permission to go through a particular door, the door still unlocks for you because you're the executive, and you're going to have a hissy fit if you're not able to go to where you're going to go to, right? That can be an AI system. I think there's a lot of really cool stuff you could do with artificial intelligence and machines and that type of deal that has nothing the fuck to do with human-order robots. And this is why I get so frustrated with Mark Zuckerberg and Metta, right? Where he is just following everybody else. And you can see how bankrupt, how bereft Mark Zuckerberg is for ideas, where basically he's just doing literally what everybody else is doing. Oh, everybody else is doing artificial intelligence. We're going to do artificial intelligence. Oh, everybody else is going to do human-order robots. He's going to do human-order robots. Oh, everybody else is going to do enterprise class AI systems. That's what we're going to do, too. And one of the things that's interesting, there's this weird blind spot. That's an important thing to understand in the technology world. That's really bizarre. This has been true. I've been in tech for 30 years. One of the most fascinating things in the technology world is so many times there are these incredibly valuable problems to solve that just nobody gives a shit about. Just nobody cares. I think about this. There was this one startup founder, and he was a pain in the ass. He was highly, he was incredibly profitable, but he just needed more money to scale out. And the problem was is he couldn't get investors to be interested in his company, even though he was highly profitable. And what it was is it was auto-routing septic trucks. So here's the thing, miles per gallon, right? You get into your car and you get 40 miles per gallon or 20 miles per gallon or whatever else. Here's the thing, septic trucks are massive trucks to begin with, and then they haul just a shit ton of liquid, just a shit ton of shit, really, right? And so when you measure miles per gallon, they're in the single digits. Like some of these septic trucks, he was talking about, they get like four or five miles per gallon. So what he looked at, right? Even though there was Google Maps out there, even though there was MapQuest back in the day and all that stuff, what he looked at is, hey, can we create a system for routing septic trucks the most efficiently? Right? So as people would call in during the day, wanting their septic systems pumped out, basically it would auto route the trucks. And simply by auto routing the trucks, it saved the companies just a tremendous amount of money. He was talking about it. He was like, he sold his subscriptions for like 50, I mean, think about this, an auto routing subscription. It was like an auto routing subscription. He sold for like $50 per month, and it was literally saving his customer something like $500 per month, just in lowering the fuel consumption for all these trucks running around. Because think about it, like a lot of these septic companies, they'll have 10 or 20 vehicles that give five miles to the gallon. Again, you start scaling, shut up, that gets to be a lot of money. And one of the interesting things talking to him is it was a bit of a pain in the ass to actually get investor money, because it wasn't cool, it wasn't cool.