Yeah, this is, to me, a very exciting and daunting as well, of course, it's hard to get one without the other. But it's my main area of research interest. And, you know, I'm all over it, that is the whole idea of getting self supervised, basically unsupervised learning. And that's what, you know, Chet GPT brought us with multimodal going from text to image to speech, you know, across the different domains. And it GPUs on Super steroid doses, as what we had in the deep learning phase. So, you know, it's very, yeah, and playing with it, you get an impression about its potency, and how it has, what we've been missing in medicine, you know, we had all these ideas about synthetic notes and keyboard liberation, and not having to deal with pre authorization, you know, automated up notes, automated bedside, you know, round notes, and we can do all that now. I mean, that's amazing. We can do all that stuff. We just need to get a chat GPT or Lowery's language model, generative AI foundation about or whatever term you like, we just have to get that to be pre trained medically. Right now, you know, what you're what we're working with, if you get on to being which is a bit a little bit more sophisticated than Chad GPT. And it'll be even more. So Chad CPT for that's coming soon. You you see the power, but it's not medically framed. Once it is, it's going to be a big, I think net positive shakeup. But it still requires humans in the loop. You know, you still have to, you know, review things. But it's really exciting. And by the way, I do want to compliment you on your team. And that grand rounds a week or two ago on this topic. Erin, and Sarah and the nominal grand rounds that if anybody's interested in a topic, it's worth spending that hour to listen, it was excellent, really well done.