Q maybe talk to why Bayesian has been coming back? Because that is even older, right? The theory of it, literally centuries old. Um, and I think when you and I talked, we, we chatted a bit about probabilistic programming. Um, is that one of the, one of the reasons, and can, can that part of the AI world see, um, Uh, rapid acceleration the way deep learning has an acceleration?
A Uh, I mean, I, I think it's partly compute. So, um, you know, finding one point, like a maxima, is a lot easier than finding lots of points in error bars. So you need more computers to, to bound uncertainty. But, you know, um, Beijing resurged in these. So in here is a Qualcomm chipset. And what it, what it does is you have a model of the transmitter that the IEEE invented. You have a model of the channel which people went out and measured. And then you receive these noisy signals on the antenna. And those, that's communications theory. It's all Bayesian. And what you do is you try to invert your model to get back to what was really said. And that's the only reason we don't drop calls. So that you would drop calls a thousand times more often if there were, if cell phone receivers didn't found uncertainty in one patient. So if you want to talk to someone, what we're really doing is just applying that communications theory to human communication instead of Wireless communication.
AI assessment note: “I think it's partly compute. So, um, you know, finding one point”