ltman has what I consider a smarter better test for AI. And he says it's when we see our first AI scientist. Meaning that the AI will discover and invent things scientifically that humans just couldn't or didn't. And once it can become like a peer of, hey, I just invented a new thing or discovered a new thing, then that would be a better test than the Turing test. I agree with him completely.

Also, interestingly, I have a dog in this race because my current strategy for survival is that I've got one more scan I have to do to see if I can qualify for a drug treatment. That's a new one that was only approved in the US in the spring. But you have to be the right kind of cancer. I have the right kind of cancer, I think. And you have to have gone through certain things that didn't work, which is now the case. The testosterone blockers worked for a little bit, but they kind of stopped working as was anticipated. We just didn't know how long it would take. Didn't take long before it stopped working. So now I'm riddled once again with tumors. But this new drug is called Pluvicto. And for some people, but not all, it can remove actually just remove all your tumors. Not for most people, but for some. It's like most things. Everybody's different. All the cancers are a little bit different. The people are a little bit different. But there's a really good chance, you know, maybe if I had to put a number on it, 30%. Something like that. 30% chance it could remove the tumors which would not remove the cancer. So I still have the cancer which means that at some rate it would return but you know maybe I could knock it back again in a few years or whatever I needed to do.

So the treatments are you go to a place and you get an IV. You go home, there's not much side effects and you do it, you know, like four to six times depending on your situation. So it's fairly civilized. You know, it's not like chemo where I'm going to wish I hadn't done it. However, it's not a cure. But if I can get this one extra scan done, it's a special scan that puts some juice in you just to find out if the Pluvicto can get to the tumors. You can't get to everything. But if it can or it can get to the tumors that matter the most, have the most lifestyle effect, then I can stall until AI gets up to speed. I do think that AI is going to cure most cancers. I do think so. Maybe not in six months, maybe not in a year, maybe in two to five. So my Hail Mary is if I can figure out how to use current technology to stay alive two years, I might, no guarantees, I might be able to bridge it to something closer to an AI treatment or an AI cure. So that's my current plan. So I have a nonzero chance of making it several years. If none of that works, if I can't get on the Pluvicto, maybe six months to a year at most, but we'll see.

Meanwhile, Microsoft is looking to start an AI healthcare service. So they're not only part owner of OpenAI and ChatGPT, but they don't want to be reliant on ChatGPT apparently for everything. So they're building their own version that they'll work into their co-pilot program and essentially try to turn it into as much of a doctor as they can. So everybody's got their own private AI doctor. Here's the problem. They want to build this thing based on the Harvard Health Publishing Arm. And maybe that's also where they're getting their most reliable healthcare information. But according to everything that I've seen about scientific studies lately, correct me if I'm wrong, but if AI trained itself on scientific studies, both existing ones that have informed what drugs are available, but also new ones that would tell us what's coming up, wouldn't it be wrong up to 50% of the time? How do you train AI to be smarter than humans when you're training it on studies that we know a full half of them are fraudulent, but we don't always know which half? Would AI know which half? Not really, because AI is only going to look at the published studies. It's not going to loo