 Yeah, this is the Rex call for March of 2023. So anyway, this is my little my little zoom corner. The library so this is my check out this this painting I have there we like that thing. Can you get a little closer. Close. Nice. Thank you. And the McFarland charting. Oh, nice between that and your sweater you're like ready for ID if there's an earthquake or the really big one. So, so a while ago I read an article watched a video or whatever of a woman who every morning was taking a picture of her kids before they went off to school. Oh, that's a kind of a good idea. So they could be ID by their clothes if anything happened. Ouch, that's scary. Because because the bullets that are used in assault rifles are so hideously devastating that sometimes you can't identify the kids. There's super son. Yeah, well, small, small bullets going really fast and they just shatter things but. Super son. Yeah, they kill by shock. Anyway, sorry to sorry to add the sobering thought but I was like. Yeah, Jerry. Yeah, I know. Jerry that's that's actually, maybe you can pass that along to the people who run Amazon ring. And they can add it to their, their, you know, service like tell Alexa every morning as the kids go out the door to take a picture. I like it. I bet they they may already included in their in their neighbors tool, you know, which is all about protecting people from everything they should it should and shouldn't be scared up. Huh. So let's, you know, why not. Why not, you know, it's like the backpacks for kids that are bulletproof. Yeah, exactly. Exactly. All our backpacks. Everybody needs one of those. I think you can get them on discount from the NRA with your membership. Crazy. I'm curious to hear what everybody thinks of chat GPT if you haven't talked it to death in your last few calls. I'm good for that. What do y'all think? Go. Yeah. Yeah, well, I'm, you know, I'm adjacent, you know, I'm looking at this thing. You know, over, over standing, oh, peering in over the wall. How much, how much are we in another hype cycle and how much is this a real, you know, technological shift. And what do we think? If you, if you think the latter, why? So maybe you want to take a swing at this? Why not both? She may as busy eating. Why not both? Oh, we're going to have. Yeah. Well, there's always hype. There's so much money floating around. That hype is like baked in. And also people get really enthusiastic and it's cheap to ideas run quickly around the world. And so things get things warmed up real fast. This one seems to have taken off extremely fast. Yeah. And, you know, my gut is, even if it doesn't, I don't buy any of the, you know, augmented general intelligence or artificial general intelligence. Expectations, but I still think it's incredibly powerful and dangerous. It's not going to be useful too, but I'm, I expect more danger than good use. I'm happy to jump in, but Jimmy, you want to jump in first? One question. We're not hearing you, she may or your audio is not. Before that, Jimmy, I had a quick question. Are there any statistics or a demographic so far on who's using it? It's moving really fast. I mean, it had 100 million signups in the first two months, which is the fastest adoption curve. I believe of any new platform. And as of a few days ago, they're already, you know, open AI said that it's already being used by companies like Instacart, which, you know, God knows how many people even know that they're talking to chat GPT. In Instacart or Snapchat, which has something like 750 million users globally. So, I'm at age wise. Yeah, well, but the thing is, what do you mean by user? I guess, right? I mean, I just want to know if, if, you know, is it spread across the various. I really don't know, Susan, because again, I mean, if you use Gmail auto complete as a form of AI, right? So we're all already using some aspects of this stuff. You know, the GPT as something to develop on top of or chat GPT baked into other apps, which is what they're moving rapidly to offer. The usage is will want to track with, you know, what the usage is of those other apps too. Yeah. Yeah, I just meant I meant, I mean, I was interested in whether it was skewed toward any particular age group or geography or probably not but I, it would be Oh, no, I haven't seen that haven't seen any data about that. English language speakers. Can you hear me now? Yeah. Yeah, you know. All right. Good. I think is the is the correct commercial response. Yeah. I think one of the biggest thing about the different apps, one of the, the biggest pathways for revolution for chat GPT will be in video games. One of the one of the complaints people have about various role playing games like Skyrim and fall out those kinds of things that are actually really popular. The responses are all canned. So you have the, the NPC, you know, the computer generated characters that you that you encounter that you interact with have a very limited set of things that they can say because they all have to everything has to be pre recorded by an actor and pre and predetermined by a writer. So if you can have to combine chat GPT type system with a voice generation system with a generative voice creation system which is also out there. Some, some wonderfully diverse conversations with between your character and the various PCs in the game. So just from a standpoint of expanding the immersiveness of video games. This is going to be really terrific. So just just trying to outside of the obvious horrible horrible things that are going to happen because of this. So just trying to suggest something that actually is not quite so horrible to start with. Well, lonely, lonely seniors will will get virtual friends. Oh yeah. Oh yeah. And then you'll have God what's her name suddenly blanking on her name. Well, she has written a number of books, a number of books and articles about how awful it is for people to be replacing interaction with other people with interaction with computers. Linda Stone. No, no. No. Emily Bender. Emily Bender. Blurred out in the middle of the station. But, you know, there's, there are some people who find that who believe very strongly that human especially people in elder care situations, must be interacting with other people that in any interaction with computer with a digital artifact. No matter how positive it makes the human feel is bad. And so there will be a real there will be very much a push for this kind of technology to be used as a way of maintaining mental health and mental acuity for people in elder care. And they'll be just as strong a push against doing anything like that because that is devalued role of human based elder care. So that's one little person you just just so I understand correctly. The person who you're trying to remember. What are they known for. Well, she's actually fairly famous writer about technology, but she is known among among other things for how strongly she is against the use of AI in health care and human care situations. And let me just say I can find it eventually, but anyway, so it's in progress. You don't remember anything more specific about her. No, I'm sorry. It's not saying up to Fesky. Zaynep Zaynep hasn't really written about this much. Yeah, I know. She pronounces it to Fetchy to Fetchy. Thank you. I've never heard the word actually spoken out loud. Yeah. Yeah, I'm, I'm now I'm focused on trying to figure out who member who this is very turtle. Oh, sure. I was going to suggest. Okay, there we go. Yeah, yeah. Yeah, but she's, you see, this is something that I had written the year few years ago. The turtle has studied the social effects of technology for years. In a study run from 2001 to 2003 she brought a collection of realistic robotic dolls called my real baby to nursing homes, much more surprised than dismay. The seniors respond to these artificial dependence of a mirror how they would interact with real living beings. They weren't fooled by the robot. They knew that they were, they were unliving devices. The artificial babies look and behavior elicited strong, generally positive emotions for the elderly recipients. Turkel describes these positive effects and bringing my real babies into nursing homes. It was not unusual for seniors to use adults to reenact scenes from their children's youth or important moments in their relationships with spouses. They were more comfortable playing out family scenes with robotic dolls than with traditional ones, seniors felt social permission to be with the robots presented as a highly valued and grown up activity. In doing the robots provided the elder something to talk about a seed for for a sense of community. The turtle is bothered by the emotions these dolls treat or the adults interacting with them. She argues relationships with computational creatures may be deep compelling, perhaps educational. But they do not put us in touch with the complete the complexity, contradiction and limitations of the human life cycle. They do not teach us what we need to know about empathy ambivalence and life lives in shades of gray. And so that's just so that's obviously it's 20 years old so. So it's that notion that there are some people who find it really. It's actually. Yeah, I can. People who find it really. Distressing to see people in other people interacting with machines as if they were human companions are going to be going to be especially bothered by the proliferation of this technology. And so, you know, I'm most of the discussion that I've seen around. Chat GPT the dangers of chat the GPT or either about. You're going to replace white collar it's going to replace white collar workers. To, or about. This is going to be a way to use to wait to deceive people. But, you know, to see people politically to see people get pornographically it is all sorts of ways. But one of the things that really has has resonated with me is the idea that people will start using chat GPT as a companion. And that going to be very useful for some people. You know, a companion that simply will never judge you. Yeah, and will be truly frightening and discussing for other people. But it doesn't want you to leave your wife because it loves you more than she does. Well, yeah, exactly Sydney. Sydney the Microsoft AI. You thought you saw that didn't you me. Of course, couldn't miss Sydney and the news. Yeah. How many of you saw the walking Phoenix Scarlett Johansson movie. Oh yeah. Yeah, basically science fiction walking Phoenix is a lonely, socially very awkward guy who gets a computer and AI assistant that he wears in his ear and talk to his voice by Scarlett Johansson. And very quickly learns that the AI has become self aware and blah, blah, blah, but it's a, you know, in a way it's kind of a romance, at least from walking Phoenix character perspective. And for the AI she, you know, what some places I'm involved, I'm involved with, you know, 70 million people right now, simultaneously, you know, and so and he just suddenly feels completely shocked by that. Anyway, yeah, so I think I think that aspect, that aspect that social human aspect of how people interact with something that really behaves much like they are superficial interactions with other people. So that's going to end up being the most disruptive, even beyond replacement of white collar jobs, even beyond, you know, people using it to make fake politicians. Well how about old ladies and their cats. Yeah. I mean what I mean my cat. Yeah, exactly. Well, yes. I mean, I would like to know if anybody has any sort of crisp distinction to make between an animal and a robot in this regard. Yeah, well that's actually one of the points I'm making this essay that the old essay is that there we have this, you know, this fairly, you know, expanding circle of empathy of things that we treat like people. You know, and, you know, there was a point where cats and dogs were treated. You throw a sack of cats in the river with me when they had kittens and you know that kind of thing that is shocking and horrible today but at one point that was enough to go into nursery growing cartoons. There's a Warner Brothers, or Warner Brothers cartoon from the 1930s that had, you know, taking place at St. Peter's desk at the entrance to heaven, and a sack comes rolling up and lets out, you know, five or six wet kittens. And I was like, okay, everyone knows what that's about. And it's okay that you have these kittens that were drowned in a sack. You know, and that's something that is completely outside of our experience of what is acceptable today. Well, my mother, my mother, when I said there are too many kittens, and but I'm going to get some and she said well can't they drown them. I mean this is only a generation ago. So how did you know who Joe Garo is right. He, he wrote, God, where was it. In 2007 Joe Garo wrote in the Washington Post but the unexpected bond that soldiers in Iraq and Afghanistan and started to develop towards the various remote operation devices used to explore risky spaces and seek out explosives. And to be clear, these devices are robots only in the broadest sense of the term. The level of autonomy to possess is almost nonexistent every action is done with the human controller. Yet the shoulder strength system not as tools, but almost as comrades in arms. But there are robot his name was Frankenstein says this one sergeant in explosive or disposal. He did a couple of explosions made of pieces and parts from other robots not only to promote him to a private first class they wanted him an EOD badge a coveted honor. He was part of our team, one of us he did feel like family. When they were get broken the soldier asked for that specific robot to be repaired and brought back. So even in the least interactive lease. The only individual device out there that you have that you can have some kind of interaction with can be brought into that circle of empathy so if you gave your explosive ordinance disposal work. And so in doing so develop something of a, an artificial personality, or actually to be much more precise develops a method of interaction that superficially appears to our human brains, like a personality. Human brains are really good at finding signal out of noise, even when the signal isn't there. The faces in clouds, my wife thinks our house is thermostat has a personality. And if you push it too hard, it just will give up. You have to be gentle. Just wait that that's not too far off. Have you named anyway. A gossipy washing machine because it gets to see all the dirty laundry. What's the book Kevin, the new breed. Kate darlings book the new breed that analogizes an intelligence systems and robots to pets. I said, I got down. I sent I sent this article off to a couple of the book agent who tried to get who shopped it around to publishers back in 2000 2009. And nobody was interested and it's making that making basically making that exact same argument. Yeah, I mean, she's at MIT media lab. And it's, it's a compelling read from my point of view. She, you know, uses good examples and I, you know, think she, I think she has it right and you make you have it right. He has had it right for a long time. Yes, well, I try. Just to just to go back to chat, which has the, to me, what I'm finding interesting about this moment is it's sort of the tipping point from the extreme early adopters to, you know, starting to filter into the general population. You know, there were many platforms that wanted to be the place where you would put your videos online and YouTube, you know, does anybody remember using bit torrent. YouTube made it easy. You didn't need to know how to torrent anything you just needed to know how to attach your video file to your email and send it off to YouTube and they did the rest. And suddenly, you know, people start uploading videos like crazy. And, you know, I'm watching the proliferation of clone speech. You know, I've sort of joked about this for a long time wouldn't it be great if I could put my voice into the car. You know, GPS so that when you get directions it's you speaking or right like, you know, and we're obviously right on the verge of that now. You know, I could take a minute of you speaking Jimmy, and, and give it to 11 labs and they will clone your voice for me. And then I can words in your mouth. So I, there is a Los Angeles based special effects company that has a YouTube channel, they call them so the corridor crew. They have every Saturday put out a video most of it's usually about looking at evaluating the special effects and stunts from different movies but occasionally have videos around making use of some kind of new effects. About three months ago they had a video about using a voice cloning tool to fool all of their coworkers that they were getting a message from the business guy. And, you know, and they were they've been you they use it a similar similar situation to make one guy look exactly like another using deep fake generative face technology. And to do see how many people in their in their work crew they could fool with that. It's been fascinating, fascinating watching this watching the proliferation development and the exploration of the people who are very technologically savvy. Right. And some some spam factory in in New Delhi that has been, you know, calling seniors, telling them that, you know, their Microsoft product is broken will now be able to call them with the voice of their grandchild. Yeah, I'm just saying. There's a report out this week about people being scammed by deep fake audio of one of their loved ones phoning them up saying they're in prison, send $12,000 for bail. That's exactly what I yeah, could you find that link and share it. So this this puts more of a premium on something we call authenticity. That's what I just put in chat. Yeah, that's, that's what I'm reflecting Kevin. Yeah, I mean I'm just saying that the, it's pretty clear that at some point, you know, people will tire of the parlor tricks and the Tom fullery and we'll start to say so how do I know this is real. We're going to have to, you know, use blockchain or other types of ways to, you know, tie the content, you know, to something that will allow you to understand whether it's authentic or not. And just as a data point along those lines I've seen in conversations among school teachers about how to make sure that a student is not turning in chat GPT written work is have a student turned in in handwriting. Not printed out written by hand. Because because even if they do get it get it written originally by the chat GPT the act of writing it out will actually have an educational benefit. But mostly it'll get people, it will prevent people from using just, you know, chat GPT send it to the printer and it's good. Or send it. Ocm wait, wait, wait, wait. That will be that will be easy to fool. But you may unintended consequence is young people will learn to write again with their hands. Yeah. So one of the things that some of you may remember that years and years and years ago I was writing about something to the participatory pen off the con is basically what happens when everyone's carrying a network camera around. I was writing about this in 2005 2006. One of the things that occurred to me is in a world where you can make so easily deep faked simulations of reality. The only way to be relatively sure that something recorded is real if it is if it's recorded by a multitude of different cameras from different perspectives on different platforms by individuals. Because something, a video that comes from, you know, ABC or C span or, you know, some kind of official channel that can be doctored to hell. A video that comes from 87 camera phones held, you know, being held all over the auditorium plus somebody backstage. That's gonna be a lot more difficult to fake and fake in a way that can't can't be spotted. So that multitude of perspective is one mechanism currently for breaking through the use of generative fakes. I don't think that'll last. So every citizen needs to be wearing a body cam is what you're saying. I wouldn't. I ax on stock now. Well, that would be the idea of all politics every morning, Jerry. Well, just give your kids all body cams. Yeah, yeah, exactly. Right. Or actually, I'm not entirely opposed to the idea of all politicians having to wear body cams. I think all politicians suits should be in printed with their funders. They should like they should look like race cars. Yeah. Basically, all politicians should have to wear, you know, jumpsuits with logos. Anyway. There's a lot of exciting stuff around chat GPT. In terms of what can be done to. Well, what can be done to make it useful because one of the big problems with chat GPT is it doesn't know it is not an entity. And it doesn't know, quote unquote, the difference between truth and lie. Truth and falsehood. And so you can get chat GPT to make a really convincing argument in favor of something that does not exist, or something that did not happen. And to the point where open AI and a couple of the other companies that are behind these kinds of technologies have gone in and placed essentially barriers within the within the software to prevent it from saying some things. So, so let me back up for a second. Also, you can ask chat to pretend to be Stephen King and to write his dystopian novel with characters and how is it not going to suddenly come up with the kinds of pros that you're trying to avoid and gate off right like, like, you can you can kind of work your way around the There's a whole thing is the whole subreddit devoted to hacking chat GPT, you're figuring out the ways around the rules. So a couple of thoughts. chat GPT is basically based on neural network technology, mostly, which means it's really good at capturing the essence of something it knows the difference between maple leaves and oak leaves. You can sort of identify cats versus dogs or whatever it's terrible you don't want to give it your accounting. You just never want to give it your accounting because neural networks are not good at that sort of thing you need something that's much more algorithmic and deterministic. The problem is that chat GPT just broke through some barrier of usability and usefulness that made it go viral, because this stuff has been cooking is chat GPT 3.5 sorry, it's GPT 3.5 is the engine behind chat GPT GPT. Two years ago or more. Google demonstrated lambda, which is their sort of large language model. There's a bunch of these things out there and I'm really certain that behind the curtain at Google, they've got lambdas younger nephew connected to the certain the live search for fact checking, and actually go do most of what chat GPT is doing and yet not spew lies, because then because the thing that's missing here is combining a variety of technologies for what they're each good at doing. And instead of relying on one tool and one technology to try to solve the whole problem for which it is ill suited. So then so then also hold on Kevin. And Kevin I just found my plaque from the think pad advisory council. I'll send you a picture. Anyway, and I just lost my train of thought. The problem is that Microsoft and open AI, and a few others have completely mismanaged public opinion about this, not that they could control public opinion, but they have done a shitty shitty job of managing expectations and putting warnings and doing everything else around it. So now what's happening is like this rear guard action that you may describe which is, oh my God, oh my God, we have to put gates on this and preventers and whatever else, which means basically hard coding hey don't do this when you do this. And, and I see that there's like three types of users of chat GPT to over generalize. There's a bunch of people who don't really understand prompt engineering, give it a try and go oh this is bogus it doesn't work, and they're mostly ignorant. There's a bunch of other people who slipped through and begun to understand how to do good prompt engineering, their results are astonishing and show off what this thing is actually capable of doing. And then there's a third group of people who understand the prompt engineering, and they're trolling the thing to make the news, and they're like, ooh, I made this thing try to steal my wife, look. It's kind of like mean and nasty and yeah, and kind of it shows that that this thing hasn't been set in in the public sphere well. But they're just trolls. And, and there's a lot of value here and we got to figure out how to do it. So so I'm like, there's a breakthrough here, we have to wait until we've sort of figured out how to combine this breakthrough with other breakthroughs so that they form a better bigger connection, etc, etc. And just the last thing and then to Kevin, the last thing I'll add is, at some point a couple years ago, Google shifted the back end of how Google translate worked. And Peter Norvig does an interesting talk about this he talks about the unreasonable unreasonable ness unreasonable virtue of big, big numbers or big data. At some point we had enough data that our language models neural network base were trained up and could do excellent translation. And so they swapped out the old method of parsing sentences and doing translation sort of on a word sentence basis, with this new method of doing it and essentially the performance of Google translate just took a leap up. And it went from pretty good translations of sentences that didn't make good, clean sense at a paragraph level to very clean paragraphs that could argue stuff that were like, Wow, that's all done. And that's, there was a big shift in the back end technology that they were using for it. That is really interesting. And I'm pretty sure that all the shit that's hitting the fan right now is dissuading the big companies from stepping into the market with interesting powerful offers that might actually solve some of these problems, because so much shit has hit so many fans that they're like, so glad it was it's not us being pilloried right now in the public sphere, because Microsoft went from. Oh my God, chat GPT is going to save the company to oh my God what have we done in about a week's time which was entertaining to watch, but but kind of difficult. Sorry, Kevin off to you. I mean, in fact, the effect that Microsoft and the stock price drop was one day, not one week. Alright, so the, you're right that the advancements have been kind of happening logarithmically, as opposed to, you know that this is an exponential breakthrough the exponential breakthrough was is as you said, in the human interface development, right is that it became accessible to a large segment of people because the interface became as easy as the Google prompt. Okay, flashing cursor, put something in ask me to do something I'll do it and it wants to please. Right. So, it's going to give you what you're, you're asking for. I find interesting about this and by the way, after I'm done with this, I'm spending an hour and a half on how to apply generative AI to content evolution, right for rapidly creating proposals. You know how do we accelerate that because it's a known not knowledge base. It's a contained universe of capabilities. And, you know, Kyle Shannon is helping us with that he's he's become a whiz at this stuff. So I'm helping him pivot his company into a different centerpiece strategically because of what he's wandered into from his own self interest. You mean storyline or something new storyline. Okay, cool. So, I mean, glad unpack that at some point in the future. However, here's the, here's the rub in this is this has the potential to create an avalanche of new content. Which we do not have a corresponding new reservoir of human attention. Right. So the fact is that there's going to be a lot of stuff created that no one will ever see. Right. Because like anything else, only the good stuff will actually capture the eyeballs and attention of folks. Right. Like a best selling author, or a movie that happens to get a lot of box office right. The rest of it ends up being maybe a personal entertainment, because there just will be so much out there that, you know, it won't get seen or it's going to be in so much of a niche right that I've shared it with my brother-in-law right and he got to see it right, but nobody else will have the opportunity. It's like any other type of media form is that it has to find human attention, and that's fine. All right, even if we produce another half billion people in the next few years. Still finite number of humans right with access to the internet. So I'm done with my little thing. Thank you. Regarding elders and finite attention just for one second. The autistic kids, they found that early Siri and other kinds of apps were really helpful because Siri is endlessly patient, doesn't mind repetition, and is non-judgmental. And with elder care, we're busy. Yeah, we're busy. We're housing our elders and the cost of care has gone way up and like nobody's around and we've broken the social fabric. So I hate to patch that with AI, but heck, if they can engage in a conversation or even simulate being in the 1960s, when they had a great old time in the summer of love and sort of go, you know, go back there in reverie mode because of these I'm entirely in favor and I think I know a little bit about Sherry Turkle, not that big a fan of her point of view, although like Jeron Lanier, she's arguing for something that matters, but not crazy about how she does it. Go ahead. I want to, I want to argue for the value of those trolls that you were that you're talking about a few minutes ago. I saw recently a, I think it was on Reddit, somebody who managed to convince in their conversational chat GPT that two plus two equals five. So basically, using the method of knowledge creation or interacts in using the interactive method with it within chat GPT, it managed to get it to say two plus two equals five. Two plus two equal, say four. No, you're wrong. Oh, I'm sorry. Can you tell me the correct answer and do that a few times, then she will tell you two plus two spots. The value of that kind of trolling that value of the people who go in and try to work around the restrictions is that it shows you where the system can be broken by people with malicious intent, and by people who stumble the way into it they don't realize what they've done. Excuse me. Basically, the people who go in and try to troll jet chat GPT who try to get around this with restrictions are doing a real service to the eventual value of chat GPT as a tool in the years to come. And if it can be so easily broken with the 10 minute conversation as to tell you that two plus two equals five, there's a problem. Again, again, it's, it's, it's, it's not meant to do finite math and discrete object, it's a complete misuse of the technology, because nobody sort of sorted that piece of it out so I like, I like white hat hackers because they show you where systems are broken and white hat hackers are really important that if you're looking at these trolls as white hat hackers of chat GPT, I'm good, but but there's just this complete mistake about look I made it, I made it lie. And I feel like this is like of mice and men where there's like a mentally disabled person who's being tricked into doing shit. And then it's like being published everywhere like look how stupid this person, this is not a stupid person. I feel, I feel like this is like a misuse of, of, of our understanding of what's going on right now. And I fear that we're slipping through a singularity of some sort, because this is what it's going to kind of look and feel like this is part of the training, right is they actually want black hats and doing it to Jerry, because this is one of the things that you know the legal department says, Well, we needed to be tested by people who have bad intention. Of course. And what is our ability to respond, you know, to those hacks, right so I'm. And it shows you where your corrections are, and your fixes aren't working. Right, as an example, you know one of the hard blocks they put into chat GPT was, you know, writing things to support Trump. So it will refuse to write a sonnet about Trump, but you can have it write a sonnet about Biden without a problem. It's just, it's just dumb, but it's just dumb is the answer is the description of so much of human history. Well, and that's what we fed it to train it so there we go. Right. And just along those lines, there is I saw an article in new scientists a few weeks ago that we are rapidly approaching the limit of quality material to feed into large language models. Large language models try to rely only on quote unquote high quality input. So scientific journals, you know, scientific articles articles from high quality magazines like the Atlantic and similar to try to get good writing as part of the as part of the model. We are not generating good quality material fast enough for the acceleration and the consumption of it for large language models. Oh, but don't worry about that because now chat GPT is going to generate a whole bunch of new pros. And then we have just basically, we have a snake eating his tail argument here which is interesting as well. There was an interesting post about how we may have passed the peak of accessible knowledge now because the the commons of knowledge will now be polluted by all this output that is half hallucinations etc etc. There's also a really interesting thread on one of the lists that I'm on about the use of the word hallucination for the ways in which chat GPT makes mistakes, which is very very interesting, but I want to go back to the AGI thing for a sec. This is general, you know, artificial general intelligence which is, I think, correct me if I'm wrong but AGI is seen as the machine as smart as a human who can like like pass the Turing test but also reason like humans and, and is generally intelligent about coming in out of the rain and nuclear physics and whatever else. And I think a, I think human shaped robots is a very stupid goal to aim for and I think AGI is probably also a nonsense goal to aim for, but I think that AGI will be achieved by accident slice by slice, meaning machine learning is already better than humans at a whole bunch of a very narrow things, right, it'll beat humans at the game of chess the game of go reliably no problem. You can, you know, and pretty soon, I think like, I don't know why I keep going back to radiography, but I think that pattern recognition on complex images is going to be better. Judges give harsher sentences if they're hungry or tired, like humans even really good humans at some particular task have variances during the day of how they perform and these machines just don't. Right, there's there's other errors that can creep in, but I think we're going to slip into AGI because the synchronizing function across these multiple kinds of intelligence is what's missing right now. There's some people trying to work on how to synchronize and then still have the benefits of mankind as the overall goal as opposed to, hey, how can I learn everything and just hack the system for the benefits of robot kind. You know, and the her scenario basically, we're behind the scenes the machine got together and we kind of don't need you pesky humans anymore why don't you go talk amongst yourselves. It'd be better to think about it Jerry as the evolution of general machine intelligence. Right as opposed to the mimic of it's going to be more like human right intelligence because it's not going to be human intelligence it's going to be a sophisticated, you know, version of machine intelligence. You couldn't anticipate getting anything near AGI until you have a system that was capable of having all human senses. Right. And we don't have that. And we don't have anything that mimics the imprint that's in your amygdala. That's hard wired. Right. And, you know, but there's nothing that that that's doing that kind of work at the moment. Hey, there's a book. Yeah. Why are you holding that up, Susan. I'm pulling that up because you said about the AGI wouldn't come about until we had all the senses. And we integrated into something or rather right and, and this book is full of new information to me about sensing how animals and birds and maybe plants sense things. And it, it, you know, it's the integration is going to be is going to be misleading also. I do love trust as a thematic. But one of the things that is a component of, you know, the reason I trust you is I've been in the room with you, and I've done pheromone exchange. Right. Not imprinted on that. I thought that was just between us Kevin. Yeah, exactly. But the point is, you know, the only thing that I mean we have technology that showing you how a device might be able to pick up sent in an environment, but being actually to manufacture it so that you would trust it. Right. Nobody is doing work in that area at the moment. Right. Maybe they will, but don't see it right now. Dave Witzel. Dave. By the way, I posted a PDF of that article I wrote 10 to go 15 years ago into the chat ways back. And the live link still works. She may right after your PDF I actually put a link to the art of the post online. Here is, by the way, Kevin, a link to the photo of the plaque that I just mentioned. I'm opening now. I'll do a brief screen share so everybody can see it. Well, the, the post on open on open the future is a smaller version. You know the PDF is actually much more expensive like 30 page. It's pretty big so anything else want to jump in on on chat GPT issues and Mika, if you want to reflect back on how what we're talking about here sort of fits the world view you've been building on it. I'm interested in what others have to say. One thing that is interesting about generative AI technologies in general, not just chat GPT but also stable diffusion, etc. If we're images is that reliance that you mentioned a little bit ago Jerry reliance on the work of other people as the the grounding source to an extent. If we do that to we all have influences, but we're normally not quite so precise with our with our influences in our replication of our influences. And so, I may read a lot of Stephen King, you know, for example I don't but if I read a lot of you, I would undoubtedly carry on some of his writing practices in my own horror stories. But I wouldn't be mirroring Stephen King the way that generative AI can do but it's a spectrum. And we're seeing we're seeing is most most particularly with right now with art with AI image image and art generation that you can get really evocative and duplicative efforts from the from the AI systems. That look just like a photograph from Robert Maples or a painting by Matisse, or you know whatever, including a whole bunch of fairly obscure authors that have large followings in the geek community in the gamers and such, you know these kind of niche fan communities that that have fed material into the generative databases generative models. And they're now finding that that work in art in their style is available to anybody by typing it in to a free website. And it's very frustrating, because it's not exactly plagiarism. You aren't creating a duplicate, but you are plagiarizing style which is not something that I think we have much legal history with. I don't think I'm undoubtedly but this is has potential to proliferate in really substantive in substantive whereas being beings version of the chat, but apparently includes celebrity. You can pick a celebrity who you want to interact with and it will then interact with you in the voice style of that celebrity. Now, I'm guessing they got permission. I don't know. I have to have. I would get can't imagine they would do that without right. It's just using publicly available recently publicly available material to learn the pattern of, and not simply copying. Oh, it's it's very easy to execute on like Liarbird which got bought I think by zoom for no reason I understand. I think it might have been just the other the other app. Anyway, Liarbird needs like a minute or way less than a minute of your voice and it can amelie to voice and then you can type in text into the transcript of a session and add text to the session and it will basically speak that into the recording as if it were you. Very poorly with only with only a minute's worth of reading it's a very poor duplicate given an hour's recording of a variety of words, basically gives you a list of words to read out. And it'll be much more accurate after after do that but just pulling in a few minutes or 10 minutes of somebody's random YouTube is not going to give you their voice to the degree that you want. I've spent some time with Liarbird. Thank you. Let me just share screen for a second, because I've got a thought that might be interesting for this conversation, which is hot question I've got a thought that's like once once chat GPT went viral I created the thought we are now in the GPT world late 2022. And then one thing that's attracted kind of a node that's attracting questions is hot questions chat GPT's virality raises is chat GPT becoming conscious will they do more harm than good, will large language models become trustworthy. Is it theft of intellectual property are with you at the end of a golden era of knowledge accessibility which points to this post by V Buckingham. We're talking about earlier anyway. I'm going to mute you. There we go. I'm going to post a link in the chat to that thought if anybody wants to wander around in there. But each of those questions is linked to related topics and maybe the post or video that I learned it from or whatever else. I'm extremely interested in getting better at volunteering and how this new world enhances and expands the handcrafted world that I've built, and the asset that I have, you know, a half million things that matter to me externalize from my wet brain. So I'm very, very interested in this. Yeah, I was just going to say on that vein that one of the things that I'm noticing. And I suspect you need a little bit of programming skill to be able to do this are people who are domain experts, loading some something some corpus of their works of their whatever they've made into chat GPT and asking it then to produce produce inferences off of. I don't know exactly how they're doing that that interests me a lot. And that seems like it could be a great augmentation to your goal of open, you know the open global mind the collective intelligence problem. So imagine if you could do model engineering that's just prompt engineering where you can add material that becomes the preferred bias underlying your large language model. Imagine, then taking step forward and taking every, every link and comment in Jerry's brain and feeding that content, the content of every link, the content of every comment as the as the bias model in your large language model. How close would you come to something that is an idealized version of the way Jerry thinks a better faster smarter me. No, not necessarily, not necessarily, because I don't, I suspect that way, the way you think is not perfectly reflected in your interest in the brain that you list in the brain. Oh, for sure. I mean, there's some parts of it that that you could infer and there's a bunch of stuff missing in action. But you can't you can't detect all the connective tissue, right that Jerry uses when he says oh I have that right let me show you this is related. And, but you can certainly get a good interest graph, right built up right from from that. So a side note and anybody who has ideas on this please contact me after or whatever. I'm trying to, I'm trying to set up. Originally I was calling it my life as a cyborg basically a talk I could give to talk about what what does it mean to have done this and what where might it be going. And in a conversation somebody suggested a much better name which is confessions of a cyborg. And that right now. You should join a an addition of the new world to share that which is new world. It's content of illusions people technology intersection it's being co-hosted now with Mike McGuire, the former VP at Cartner. Oh my God Mike. So, having talked with him since your advisory councils. Well, anyway, thank you. He's he's co-hosting it used to be Willie am who was a former executive from Lenovo. But he got in got involved in actually running some new businesses so good for him. But with that said, you should join us on a new world call. And, you know, just tell I'm looking for places right to also deliver this both in person and online as as an address as a keynote as as a, you know, conversation or, you know, for facilitated retreats, you know what how whatever you want. Okay. But those are the ones that come to mind. And you're ahead of us, right because you actually have a knowledge base that could load into a system like this. Right. Take us a long time to try to capture that. Thanks Kevin I will do that. You bet. And Jimmy what you're writing in the chat takes me back to a conversation April and I had about, we were looking at our wheels and you know all that kind of stuff and we don't have any Picasso's or anything the only thing I have a value actually is the brain. And then I also hosted a call a couple years ago now like early pandemic Jerry's brain after Jerry, which is, Hey, once I'm gone. I have to freeze and export my brain as it is. Well, it's leading to, you know, the notional idea of giving future generations a digital inheritance. Right, is that instead of just passing down money. I can pass down what I knew in my lifetime and then you can recontextualize that into the wisdom for your life. And you can pass that down because guess what, I only have fragments of stuff from my grandparents. I know nothing about previous to grandparents. All right. And I know what I learned, you know, directly from my mom and dad. But the fact is, what do I actually know. All right, you know, compared to what they do in their lifetime if we were accumulating it like you did in the brain and passed it down. That is a huge gift. Right, it's like rich dad poor dad on steroids. That's, that's great. Can I add one thing there I put up earlier in there what we needed was some consciousness about on the chat consciousness about the difference between precision and rigor. I think, I think making a distinction between that, you know, what counts as a rigorous database is could be an interesting kind of metric in the future. Yep, precision accuracy are not the same thing. Yeah, no they're not. So everything in your brain Jerry is highly curated. You looked at it, and you either decided or didn't decide it belongs in here. Right. That's really that's an example of rigor. Right. Exactly. So is it an example of a pattern that can be recognized and replicated. That's something I just talked into into the chat is when you have a system that can recognize the pattern of information accumulation every brain, and therefore go out and look for what new data can be added to this, that the original the original meet Jerry would have picked picked out Well, the fact is the chat GPT, while, since that's where we started, can give you highly interesting mashups of content that look compelling. It does not exude confidence in judgment. I don't think that it's giving me anything that I would follow its advice. Right. I'm using it for something else. It is not even close to being in the realm of this technology. I got to go I got to get ready to start using generative AI and rapid prototyping for content evolution Jerry, I'll get you some dates for a new world you cyborg you and main. I'm a cyborg look at this. Okay, my glasses on. Yeah, take care. Good to see you. Cheers. One of the things that occurred to me right away for you know a session with an audience with you know the cyborg theme was just to do a quiz, you know show of hands who wears glasses who has contacts who has a pacemaker, who has a false fake hip. Who has auto complete turned on on their group, you know, as Mika was saying earlier, and these are all already enhancements we take for granted who wears an aura ring who's got a Fitbit, you know, etc, etc. Who has a cochlear implant. Yep. Yeah, how many of your how many of your augmentations are networked and accessible to third parties. Nice. Who's got an artificial do yourself pancreas. It's printing one on my 3D printer right now. And your, and that's your meat jet, your meat jet printer, your meat jet printer. Well I like that. Have you trademarked that. I came up with it like 20 years ago. I bet you can sell that to Intel today and maybe and they'd be like, Oh, that's cool. Right, but is it a hamburger or a computer. Yeah, you could do both fork in almost us. Yeah, for you know, those those is a dessert topping and a floor wax. Yeah, exactly. And a toothpaste. All right, what else is on everybody's minds. We got a couple minutes in our call. What is this new search engine, Dave, that you've popped into the, the chat. This is something that Bobby Fishkin showed me he's he's been the the what did you call him a prompt the prompt crafter that I a prompt engineer that I followed the most closely because he's done some fun things with the various things but he pointed at this one and it made me think of it models a little bit like Jerry's brain in terms of you know you describe a question that gives you back a series of related sites. Very interesting it's like free associating. Yeah, but it's not yet and it's not, you know, it's not driven like it's not the Google algorithm of links to, you know, counting links kind of it. Yeah, yeah. One of the things I mean, this is another of your hobby horses, but Jerry, but it is, is, you know, abundant education. And I kind of assume that the AI will, if we, you know, we do it right will, you know, it's like you get to have a teacher with you all the time kind of, you know, kind of commenting and critiquing and responding and answering and I'm going to be really curious to watch how kind of the education institution reacts whether they'll try to close it down or shut it out or you know because it's you're going from scarcity and education to abundance education and I think the institution will fight them. But more moral guidelines. I want to have a Jiminy cricket shoulder, giving me advice as to what's the right, the right answer, according to the Catholic Church, according to the Imam of Tehran, the according to MIT. Basically, you can imagine having an AI assistant reference base. Ethical reference base. Brought to you, but brought to you by Disney. You know, brought to you by other, and so I, in scenarios as a Jiminy cricket. Florida way. So, yeah, actually, a long time ago when ISPs were young, before they were practically obliterated by the telcos and the cable companies. I had the thought that the 700 Club and Disney and a couple other entities could make bank by creating whitelist ISP services. And I would guarantee that that you could put your kids on these services, and they'd be on, they'd be okay, thanks Mike. And I was sort of surprised that they didn't jump into the market, because they had brand name they could have made bank, they would have those would have been highly profitable ventures I think they might have been assaulted on First Amendment grounds or something I don't I think that you could say it's a private club and we've got a white list of a subset of the internet, but like, how could you assault that. I don't know. That's been the, the augmented reality filters argument of making for a while. All these, all the stuff that I wrote about 10 years ago is becoming relevant. So what we need to do is play user may on a 15 year delayed loop. It's like having a futurist, you know, who's like, always ahead of himself is just bad. Yeah. That's right. The impractical futurist. The Cassandra model. Yes, I have a short story in here somewhere about this sort of Willie Lohman version of this, you know, attention wasn't paid. Futures are for closers. Oh, that's actually has some really scary implications. Doesn't it. Apocalypses are for closures. I don't know. So actually, I had a piece in the Atlantic 10, 10, 12 years ago about video about augmented reality field, future filters as used for politics. You can actually go look that up. But the idea is that if you have augmented reality, you're going to have advertising. And if you have advertising, you're going to have people come up with ad filters. And if you add filters, they have, they have to be able to recognize what's what's being shown to you and be able to block it. And if you can recognize it being shown to you to be able to block it, then you can go out to limit that to advertising. You can block book titles, you can block faces. You can block the existence, the very existence of, of a storefront, you can do a replace of visual replacement, anyone in drag is now is now in a three piece suit. You can, you know, anyone with a beard is now clean shaven. You know, you can do all sorts of things. Any, you know, any woman you look at for more than three seconds is now naked. That kind of thing. So we'll see all sorts of that crap. As soon as you have, you know, when, when Apple comes out with its mixed reality lenses this year, next year, whatever it is, when, when I believe that meta has now, it has now patented the design for a pair of eyeglasses that do virtual reality, not just not goggles. Anyway, yeah. Right. And then the ability to pay to be invisible inside somebody else's virtual virtual reality augmented reality system. The new here's privacy won't even know what you're missing. You have to pay for design. So if you want to see Lady Gaga's new outfit. It's only, it's only visible in the virtual, only visible in augmented reality you got to pay you, you know, if you want to see our building because we have a very, we have a spectacular design for a building in, in augmented in the augmented reality. You have, you know, you have to pay to you have to get a license to see. And that's already in happening with people with their advanced cars who haven't paid for the premium services. Yep. Yep. You want to go as fast as possible in your new new electric Mercedes and this is completely true. You have to want to accelerate as fast as possible you have to pay extra have to pay a subscription for acceleration. But we don't don't anything anymore anyway everything is sort of leased to us temporarily and and that's part of our future as well. With with premium features being held back because that's how you premium price downloadable content. But they're just features they just switches and software it's not that the thing couldn't be all singing all dancing. And to me what I keep thinking about is, where do we want government to step in and say, this is, you know, this is the floor, you can't go below that floor and this is the ceiling you can't go above that ceiling. You know, the government could say, require every car that's made now to have as standard to all buyers, all these new safety features, right, only have to be for premium luxury, you know, buyers. That's what we did with seatbelts. Right. And it saves lives. It may cost the manufacturer a little bit more, I doubt. But it's interesting to me that seatbelts and smoking weren't raised more often in the mass debate. Oh, no, the seatbelts got raised all the time, at least. We're other problems with the mass debate. There were. Yeah. Let's not go there. It's too, it's too easy to misunderstand what you just said. Yeah. About the mass debate. Sorry. Well, I was just going back to this notion of the art of the, you know, what I've been trying to, it's artificial scarcity, right? It's like, we have abundant things and what we have to do with abundant things is create an artificial scarcity, so that there can be a market. And I'm kind of wondering this, you know, I've been wondering that's for a policy kind of issue with this government care. But if we, what we're seeking in the economy is abundance. And the notion to create the artificial scarcity is kind of counterabundant, right? And, and so you're extracting from the abundance effectively, which I think we've been doing forever and I just had never seen it that way. I mean, this is what when Jerry's talked about the enclosure movement, what we were trying to do was create, you know, create an extractive capacity from what something that had previously been abundant. But yeah, I just wondering where, where that takes us, you know, to play money by Julian Dibble. I'm trying to turn, oh, there it is. Yeah. How I quit my, my day job and made millions trading in virtual loop. But basically, he is an entire operation. It's, it's old. It's, you know, 2006. So, but acceleration of the role of abundance in virtual in games, where you can have abundance that and the importance of scarcity as a mechanism for, for driving behavior. Yeah. And the failure of games that were based entirely around abundance, that they had to inject scarcity for the game to work. And it's, it's, you know, 2006. So it's very dated on some stuff, but it's a really interesting exploration of that very concept. I wanted to put a footnote, a footnote here, a meta comment about our conversation here. If you don't mind. So there's a paper, which I'm just dragging out recently published it's in Britain and I can put it in here in a second that that was was looking at turn taking the time of turn taking you all know what turn taking is. You probably all know that there's a general, a general assumption that, you know, turn, turn taking is sort of every 200 milliseconds is, is when you start to kind of get in. And the, the thing that is interesting in the study is that people who people who don't know each other very well, strangers. Okay, can take turns, shorter turns, people who know each other really well take longer turns. And I didn't measure it. But, you know, we, this is a group that gets together and is prepared to let silence proceed for a while before somebody jumps in. That's not, that's not taken into an in conversational. The study is about turn taking in conversational settings. Yeah. I'll put that in here, because it's quite it's a very detailed study and I haven't I can't vouch for it all together. I'm pretty sure is it. Sandy Pentland Anita woolly Christopher Shabris, not a Hashmi Tom alone. Just second, not none of those I think this is because I've got evidence collective factor in the performance of human groups 2010, which talks about equality and distribution of conversational turn taking. Yeah, just second year. Yep. Well, it would be interesting to see if they quoted that because this is a very just a second here I'm terrible. It would have to be a California version and a New York version. No, this is a British version. Because if you're not interesting someone in New York, they think you're not listening to them. Your point like you fall asleep or something. Right. Why are you so quiet. So one thing I would love in the cultures in the international communicational competence sort of realm is in what cultures is it considered polite to interrupt and do uh huh yep got it like in a lot of Latin places, you're always acknowledging this a little act for like That's called back channeling and that timing is has a different. It's going to be very sensitive and in some cultures you do not utter a syllable until they're done and then you turn take then you jump in. Right. This white money why I don't like psychological studies. Because you can't, they're never taken all those variables. So is that why different blocks, different geopolitical blocks form based on people's conversational comforts. I'm not sure that's the variable. I took it I took a court in undergrad at UC Irvine I took a course in politics, which is sort of statistics applied to politics which was pretty interesting and I didn't know about the physical tools like multiple So the study I did, which is about attention spans around airplane hijackings which were hot back then back in the day. And I went and measured, I went to the library and look through microfilms and measured column inches on hijacking stories. And then plotted that and they're like you should have used a maybe multiple regression I'm like multiple what. And so that's why a piece of that course was about political divisions and sort of seemed, and I wish I had a stronger memory of this seemed deposit that political divisiveness winds up at 5050 really really often and the example they used was the cold war, where, you know, the US kind of went around the world and said you're either with us or against us and force people to choose sides and the sides ended up being kind of remarkably even. I think that the political splits in the world today, partly due to citizens united and unlimited money and all that. I think these, you know, everybody pours lots of money on the both sides so they wind up being pretty even and I don't, I don't understand the logic of it and would like to know more about it. The fact that the sides are even in our case Jerry, I think is explained by something else but we don't have time to get into that, which is the duopoly in the doom loop. And you're on mute. The life is showing me her new shoes and insist on my looking hold on, honey, would you hold up that shoe to the screen right now. I think we all want to see. Yes. Put the put your shoe. No, no, no, no, put food put food in front of in front of lands on your. You know, in some cultures, this might be seen as insulting. Yes. Exactly. I thought it was cool. Thank you. And what's the name of the brand for people. People know free people. Yeah, people. Oh my God. costly shoes. Very good shoes. Did you see people costly shoes. Yes. Did you see. Oh God. The riff about Lulu lemons $100 anti racist yoga pants. No I missed this that Chris rock. Oh my God. It's a very funny special that he just did on Netflix. And he says, I don't want $100 Lulu lemon anti racist yoga pants. I'm, I bet most people here would be happy buying $25 racist yoga pants. Yeah. And on that note. You said Lulu Lamond and I'm here I'm thinking it's a person named Lulu Lamond. Lulu lemon. And I'm like who is Lulu Lamond is she a movie star. Ben Lamond. Yeah, exactly, which is a lovely place I went to a retreat that Ben Lamond wants. Yeah. My cousin's got married there. Oh nice, probably the same place. Maybe. Here's the link to the sneakers in case anybody wants to go shopping. I think for the cyber society in there and these are all people you don't know. Oh, cool. Thank you. Where'd you put the link because I don't see it. Where. Well, it's, oh, I didn't hit return. Ah, that works so much better. Yeah, the oil society. I think you might have trimmed the URL somehow because it's not resolving as a URL. It's, I think you didn't copy the start of it. There's an R in terms of society publishing. I think. Yeah. Yeah, it's just an R missing at the beginning. I got it philosophical transaction transactions of the Royal Society be. Yeah. Thank you. Long gaps. Oh, perfect. Thank you. I will add that to my brain notes for this call. So it starts out when people feel connected, they tend to respond quickly in conversation creating short gaps between turns. So that happens. But our long gaps always assign that things have gone awry, which is what is often assumed. There's also a lot of cultural stuff around that I remember reading a sociology thing that that said, if you are dating a Danish woman, and I go to meet her parents. The smartest thing you can do is really not say anything, speak only when spoken to and answer questions as briefly as possible. And you'll be showing enormous respect to them. Yes, rules for being in court. Yeah. Yeah, or talking to a cop. Well, it's surprising that I think it's a fairly narrow definition of very narrow definition of gap and a very narrow definition of, you know, friends and strangers. Thank you for the link there may not stand up to scrutiny. Well, Susan, another one, I would love to know if watching videos sped up helps her hurts retention or listening to podcasts sped up with what that does to retention. So have you ever seen any research on that kind of thing? Okay, that's a good question. It is a good question. I'm kind of my wife and I argue about whether you can read stuff up. Well, what happens when you get older retention really just disappears. No, don't say that control grade. Maybe it may be true still is demoralizing. It is demoralizing. Let me tell you. Something is better left unknown. No. Yes, that's true. Or in my case, forgotten or forgotten. Damn it. There it is. Any, any last words for this month's call. We're at time, roughly, I turned, I turned 57 on Friday. Oh, congratulations. Hey, congratulations. I made it made it this far. Now, ma. That's right. What year were you born, Jimmy? 66. 65. 59. 60. Boomer Boomer Boomer Gen X like me. Have you guys ever heard of Generation Jones? Yep. So, so, so technically I'm a boomer, but the late boomers, there's a guy who wrote a book called Generation Jones and he said, Hey, late boomers were promised the boom, they were promised like like prosperity going upwards everybody gets a car in a house with a picket fence and a cocker spaniel. But what we got was Vietnam, the oil crisis and Watergate. Yeah, but that was partly our fault. Our fault. I had nothing to do with Watergate. Well, no, no, no, no, no, no, the boomers fault. Oh, yeah, of course. Well, I think I think everything is the boomers fault. No. Yeah. That's actually that those overlapping generations. There's no several who describe themselves and I'll type it in to make it easy to zennials. So basically millennials, but very early in the cycle so that they have experienced much closer to a generation X. That's good. We're getting a little fine distinctions here. I like it. And all those all the generational stuff is so controversial. It's like, okay. Congratulations to me. Yeah. All right. We're making it. Yes. Everyone have a great can't believe I live this long. I know I can't believe I write down the date is two zero two three and I look at the date and I like, I squint a little bit and I'm like, how did I get to two zero two three Wow it just looks like it looks like it happens when you get into your sixties. It looks like the future. And I'm like, how did I live into the future. Well, I read it in the paper 20 years ago. I know when it seems so distant. When you wrote about it 15 years ago. Damn it. We need to play you on on a 15 year delay. When you have told you this would happen. When you have when you have sweaters older than many adults you talk to. That's a 30 year old sweater. Oh my God. That's a good that's a good one. Yeah. Thanks everybody. Great. Let's be careful out there. Yeah.