What really is AGI and how close are we to it?

Thing is, I wonder if AGI is in the form that we interact with LLMs right now. I think it has to be something else. Um, I don't think it is a... really... well, I guess it could be a really smart LLM model, but it has to—I feel like it has to be able to have more physical capabilities. I think it has to be an interaction between perfect sync between amazing model and a hardware that is perfectly made to match this. Um, and so the idea is like, obviously, once you get to AGI, that's a model that can do anything basically, that can... is smarter than a human. And obviously, models are right now, they are probably smarter than most humans, but... but there is little things that will trip up on. They cannot help you bring water from the kitchen and like do more physical... um, physical work things. But there's a general model that has like general intelligence and anything and everything. For example, the way models trip up on spelling like strawberry... yeah, like they trip up because of how LLMs are trained. Um, but a model that actually knows everything, that won't trip up on silly questions, um, I think it also needs to have some physical capabilities to be able to actually like interact and do real world things, 'cause there's some things that you can only answer properly or like explain properly if you're able to feel and touch. Okay, what exactly is that? You want to ask it, what's that, and it should be able to like feel it.