There are related concepts, but Moravec's paradox is the sort of imbalance. But the Centaur thing is the sort of synthesis of that, where it's like, okay, like human plus AI outperforms either one of those things. And actually not true in gaming, but let's leave that aside. I do think like, yeah, so I think like this kind of debate, which by the way, I have not heard the same data points that you have, but I do agree that radiologists has not died as a profession. I think it is a fundamental discussion about what a job is, whether it is the sort of specific object recognition. So for radiology, it was like, look at this scan and determine if I have cancer or whatever. That kind of thing can be automated, but it's almost like you can give me AI assistance for that based on normal cases, but without any context of the patient history or any understanding of the flaws in the model, which again, because the model has not solved continual learning, it does not learn. And so it would just repeat the same flaws over and over again. And it takes a human to catch those flaws. And it turns out when it comes to health, we really care about the flaws. So yes, radiology has not died. But I think it's the same thing as this whole discussion about is AI going to take away software engineer jobs or is AI going to kill software as a service companies and the entire industry in general. It turns out that we strongly, as software engineers, we strongly identify by typing character by character into our IDEs. And now we no longer do. But I don't feel threatened with my job. I don't think you're threatened in your job because we still know how it works and we still know how to use AI to do our jobs. We'll probably be more effective at it. And in the same way for radiologists, did they strongly identify with being the best at pouring over those scans and calling out those things? Yeah, sure. But once AI was better than humans, like measurably better than humans at it, then they moved on to other parts. They just move up layers of extraction. And I guarantee you, humans have no end to the amount of extractions we can make up once a thing is solved. And also those abstractions come with costs in the same way that automatic memory management in programming languages has a cost that we usually eat. We usually pay for, like, we're like, happily, we'll pay for it. Right. Sometimes we want to go down a level. Regardless, like the job is not threatened. Like we're using better, better, better tools. I do think like there are a lot of, so the counterpoint is that there are a lot of bullshit jobs. You know, just last week, a block slash goose laid off 40% of their workforce. And I don't know. Stock went up, and like everyone was like, Yeah, those people weren't working very hard anyway. And so, like, yes, there's uh, people who are working bullshit jobs should feel afraid for their jobs, but you knew you were working a bullshit job, you knew you weren't working that hard, you knew this, right? So, like, this is not this is not something that's like existential about the discipline, it's just that you were underutilized. And guess what? The economy is coming for you, and you will be put to better use elsewhere, or you're just not going to work. And, like, isn't that nice? Like, like, uh, you know, the ultimate reality of abundance is that everyone just has a basic living wage, uh, universal basic income, and they can live their lives, have fun with their kids, touch grass, while the robots work for you. Why are you so scared of this?

Yeah, I mean, that is certainly one way it could play out. And then there's also, you know, the Skynet kills everybody, you know, and we're all paperclips. I do, yeah.

Yeah, I have an unusual take on that, maybe, which is not something I've heard other people say, which is that it seems to me that the history of human disasters is that the way that we prevent mega disasters is that we have a smaller disaster that's still really bad. Yes.

But

not a complete Armageddon. Where does