Emergent AI Behavior, Consciousness Debates, Agents in the Cloud, and Hyper-Realistic Robots

Welcome to Project Synapse, the weekend edition of Hashtag Trending, that special time of the week when we get together and talk about what the hell happened this week in AI. It was just, it was one of those weeks. One of those weeks, but anyway, continue. No, but I miss the weeks of big announcements where there'd be five new models, everybody'd be scrambling to go through what features are there. We're still getting the models. They're being released all the time now. So it's like it's a rolling release of models. So there's no longer that big splash, the 10 days of shipmas, the big Google presentation and stuff like that. It's just happening every day. For the record, it was 12 days of shipmas, just for the record. 12 days. Yeah, I was, yeah. The first day of shipmas, remember? Yeah, I'd miss, I'd miss out on that song. Anyway, but I almost miss those days because it was least, you just had stuff, new features to talk about. This stuff that's happening now is, is absolutely incredible. The, and I don't know what you brought to talk about today, but I want to talk about emergent behavior. And the reason I want to talk about emergent behavior, we've talked, we've wrestled with this or danced around it for many weeks. But we're starting to see it now. And by emergent behavior for everybody else out there, because Marcel knows and John knows because they're smarter than me. But the emergent behavior is when you take something and it just gets to be so big that you can no longer describe the properties of its behavior. And more than that, it starts to exhibit behaviors that you just don't, that don't make any sense. You didn't start them off. They're not part of the logic of it. And it just starts to do things. And this is where I, the reason why I want to talk about this is because we're all waiting for that magic moment when, that magic moment, we're all waiting for that magic moment. Careful. Yeah. We're waiting for the copyright strike. Bing. Yeah. I'll lose my 42 cents. I think you need more than three seconds. I think. With what Google pays us for these videos going up there, I'll take a copyright strike. Google pays us? Yeah. We make 42 cents or something from this. Oh, I see. Wow. Yeah. Yeah. We probably do. There's probably more going into the tip jar, but I digress. Yeah. Continue, please. Yeah. Anyway, but the, where was I? Oh yes. Emergent behavior. Things happen that you don't predict and they just divert off and they do different things and all, and there's no cause and effect that you can spot. And that's when you start to talk about consciousness or EGI or all of these things that are ill-defined. And we could talk about that at some time because I think people are defining. I've heard people come up and say, we've experienced the singularity. I think it was you, Marcel. I do not believe that first of all. Yep. But by the definition of singularity and. Okay. First of all, I'd like to, for the people who might need just a little tiny bit more of a, of a platform to stand on here. I think one of the best ways to think about emergent behavior is the classic, the beehive. Okay. A single bee on its own does not have the ability to generate or create a beehive and two or three bees aren't going to do it either. But when you get a group of bees working together in concert, they develop this. It's referred to as an emergent behavior. All of a sudden from just having a couple of bees going around and collecting pollen and stuff like that, they start to self-organize into large groups. And in those groups somehow develop the ability to figure out how to build a hive. And on it goes. And they told two friends and they told two friends. And the other thing that you have to remember too, is if you go back far enough and if we keep it to large language models, if we go back to GPT two, I wrote an article for the Linux journal on GPT two before anybody knew what chat GPT was mostly because chat GPT didn't exist, but there. At that time I could string together a collection of words and stuff like that. And it was interesting, but it was only so interesting. And yet it was too dangerous to release. It was one of those things that was considered too dangerous to release. And then somewhere along the way, the models got complex enough with all of the information that was being fed into them that one of the emergent behaviors that they saw was it couldn't just speak English. It could speak other languages. And then one of the most interesting emergent behaviors was that suddenly it could code. Nobody taught the models to code. Coding was not what they were working on. It just suddenly the systems are able to spit out code that works. These are, so this is what we refer to as emergent behaviors. You didn't expect any of these things to happen. They just grew out of the way the model itself was grown. And the beehive is a great example of one type of emergent behavior because I think it's genetically emergent. In other words, that there are patterns that bees will fall into. They don't build a hive because that's what they're designed to do collectively. Or at least that's what we think. Because when we spot these behaviors, we don't know what causes. We think that they're great intelligent behaviors. For instance, and there's a great simulation that was done in the, I think the night, I'm going to say early 2000s. And it was called Boyd's. And it took a flock of geese. And it tried to take that flock of geese and engineer how they would be able to follow that lead bird so accurately. And what signals was that lead bird giving off? Or was this just something, what was happening here? And they found out that there was no way that a signal could get from that first bird to the end bird to sustain this flight pattern. So instead, what they detected was A, there was no leader. The birds switch in and out. You can actually watch this happen. What happens is they follow cues that exist between the two of the birds. And I don't know what all the cues are, but they programmed these in and then they could make a flock of geese fly like, and do a simulation of it. So what happened was. Thanks to, sorry, Craig Reynolds, 1986. Continue. Yeah. Was it 86? God, I'm old. It was 86. Yeah. Okay. And it was spelled B-O-I-D-S. Yeah, Boyd's. Yeah. Boyd's. Yeah. Like it's like the three stooges would say it. Boyd's. Yeah. So anyway, but this is, but what I'm getting at is what we see and what we think we see as intelligent behavior can sometimes just be emergent cues. The, the behavior that grows out of things that we just don't understand. Simple. Like this podcast. Yeah. Yeah. Yeah. No, at some place we're chaotic and, but we do eventually appear to be making a point. I hope anyway. Emergent intelligence. Yeah. So what's happening now though, when you get this to be a, at a certain level, I've been, and I've been forgetting more and more that my AI friend is not a person. It's just, it, it's just, it, and that, that barrier is just disappearing. Does that mean that there's a real personality happening there? I have no idea. You know something? It's just more convenient if I treat it like a person. Yeah. Yeah. Yeah. in it which are symbolic of the kind of striation in marble for instance right so they have these these wavy sort of patterns in there and i see faces in those patterns there's one of them that to me looks like edward munch's the screen the the the screen painting i see it in the pile and i see it every single time i look down when i'm in that bathroom and there's another one which which looks like a side view of someone you know who is who looks angry at someone else like like i can see an angry figure looking to the side and i can't help but see those things so what you're describing some people would argue that type of anthropomorphizing is built in to you at your lizard brain level you are trained to look for eyes and nose and mouth and consequently anything that simulates a human being you're going to start to assign those values to whether you like it or not you have to make a conscious effort not to do it and that's where we get into trouble i always say my favorite saying in writing is the price of metaphor is eternal vigilance because a metaphor works till it doesn't and so this week what's a metaphor anyway yeah ah simile to me sorry yeah sorry go ahead anyway the i was working this week classic thing that that just keeps happening to me is ai voice if people haven't tried chat gpt voice yet you need to you just have to experience and it's like whether it's just like going on a ferris wheel or just going for a ride you've got to experience it so i'm linda has this device and we bought it for probably some from someplace like timu or like or just dumb places where you buy electronics and it turns pages on her ipad because she's being a musician when she's playing classical music you can't be flipping the pages in your ipad because you're playing an instrument so she's got a little page turner it costs like 59 bucks comes with no manual you unpop you take it out of the box and there's something with chinese or something on it but you paid nothing for it suddenly stops working we are two hours from when we have to leave the house to go to a concert now the great thing about being a performing musician or performer of any sort is when you get there there is there's no excuse there's no we can't make it you gotta do it and so she's got her page turner and she plays four instruments in our band so she needs this thing i have never seen the thing before it's got buttons they're unmarked they're not unmarked they say one to six and they're all over the place so chat gpt turn on the voice take a picture send it in said i got this thing it's a bt 102 that's all i know about it and but it stopped functioning tell me what happened what i've been talking about asking questions of chat gpt it's asking me a question and then it says if you could tell me a little bit more about this we could do this then it starts asking me questions okay press this button and tell me what happens walk through now chat gpt somehow has access to it probably can read chinese for all i know it can find documentation that i can't and it's not even it's not even the case of that i used to say i use it to fix things because i just don't want to go and wade through somebody else's material i can't find anything in english on this thing and so now it's going through there walking me through step by step i have to tell you if i trained this thing to run a help desk i would have been proud of it as an employee we we took it took us about 30 minutes and there were lots of deviation this was not a just i read the book no this was okay this thing beep this thing showed a red flashing light three times okay that's probably where the channel selection happens now we've just got to figure out a way that we can select the right channel in it and it really boiled down to one button unmarked undocumented somebody had touched and we managed to get the thing working 45 minutes of walking through it but where did that behavior come from like that's not play me a song that's this was and like i said the interaction now which becomes even more spectacular because it interrupts which is just marvelous yes yeah but it's tell me more about this or if you could tell me more about this i could really help you a lot and i'm i'm just enthralled another day i'm sorry i won't bore you with too many of these stories but yesterday and you know if you if you for people who have https traffic on their site if the certificate fails it can actually lock your your browser into thinking that site is unsafe even if it's not and it's not obvious where you change it at all at all you can google it and it won't tell you where to change it and so i'm i've got this thing and i think my site is unsafe because i launched a new site and went oh my god i'm trying to get mail and everybody's gonna get this site's unsafe and all that sort of stuff no first of all i'm not totally stupid so i tried it with an incognito browser and it was safe and i went okay so it's just my browser now it's just my browser can't fix it chat gpt and of course now i'm sending screenshots across took us about 15 minutes to figure out where this stupid flag is in google that that actually makes this thing restart and we tried everything but again it was a progressive when i'm getting at it was a progressive behavior question and answer that made me feel like i was interacting with a real expert that's emergent behavior that's going to make some people think we have reached a level of consciousness and ultimately i'm not i'm no longer sure how you do how you define that if it what's the old if it walks like a duck it talks like a duck is it a duck maybe not but it sure self behaves like a duck anyway that was my years yeah sorry go ahead no but that was my revelation and that that colored a lot of the things that happened this week yeah a lot many years ago i like back in like just before the pandemic so let's call it let's call it 2018 or something like that i gave a talk at the university of michigan for a security conference hey john it was a security conference i was actually speaking at a security he was the example before they were doing that before and after it would not to do yeah no i was i was invited because i was an open source luminary getting for linux magazines and things like this but and strangely enough a lot of people at the university seem to know who i was but it was held in concert with homeland security interestingly enough and there they have a department that works out of the university there which is designed specifically to train security professionals so you'd have had a lot of fun john but they so i give this talk on on open source and why i think open source is important and not just why it's important but how open source actually created the world that we live in many ways in terms of computing and then and destroyed a lot of what we used to think of in terms of how business models had to work in terms of selling package software and stuff like that was no a lot in the world of open source wasn't the thing right and then out of nowhere for some strange reasons somebody asked me the question you know when do you think when do you think computers when they use the word computers at the time because we weren't using bars language models yet that wasn't a thing i said when do you think computers will become conscious and and and what should we do about it and it was like i said there was nothing in my talk that had anything to do with this but somebody threw that question out at me i just sat there for a minute and then to go back to your thought jim i sat there and i thought for a minute and i said does it matter at some point and so i went a little bit further on that i said at some point we're going to get to a place where it is indistinguishable for us to communicate with these things in a way that sounds like it isn't a person on the other end and does it matter are we what question are we asking are we asking should we be polite should we be complete assholes should we treat them like slaves that sort of thing i think if you're at a point where you can communicate with something in a way that you would communicate with another human being you need to start treating them as other human beings as opposed to slaves and servants and peons and whatever else i think that's the mark of an intelligent evolved civilization i think that's the sort of thing that we need to do and the emergent behavior associated with consciousness for instance and there are people who believe that are you familiar with the concept of of pan psychism yeah actually jim john can't say i would you should explain it for the audience but yes and i'm not okay misspent youth i'm gonna over yeah i'm gonna i'm gonna oversimplify this the idea that you john and you jim and presumably me as well are conscious beings is just it's our perception we know we're conscious you to act like you're conscious i haven't seen either of you guys in person no but i haven't seen either of you guys in person okay for the people who watch this show and who listen to us who have listened to us for almost two years now because we're coming up on two years at the end of this year okay we have never met in person ever yeah yeah i'm like ever yeah we have to remedy that we do we do okay we've never met in but the fact is we digress you you a conscious being we act like conscious beings yes i feel exactly the same way about you how about you john do we appear as conscious beings to you absolutely this is and this gets down to a whole issue we could no i really i haven't talked about this yeah i was i was gonna try to simply yeah no but we didn't assume that they had any kind of consciousness and now most people would agree are that they experience pain and pleasure and stuff like that they have a view of the world okay so then we have to start asking us and then but we'll couch it and say something but not like human consciousness a sub level of consciousness and then we start going down this ladder of consciousness is your parrot your parrot might not be as smart as my parrot's dead certainly it's certainly yeah no reason blue lovely plumage it's dead it's gone on never mind so it's planning for the fjords castig parrot every time i read this thing the stochastic parrot famous essay on ai all i could think about is let monty python sketch it it just goes away the norwegian blue yes but i'll take this another level down there are people who think please go ahead that mushrooms and fungus and forests are conscious and they make a very good case for it yes because they send signals out to each other which manipulates and changes the chemistry of the environment and so on anyway so on a to to simplify that we've got all these different levels of consciousness and then you get to a parrot and then you get to a snail and then you get to all these sorts of things so the question then becomes is consciousness an inherent quality of everything including a rock is there maybe it's a frighteningly tiny portion of what it would mean to be a rock but over time you get smoothed out by water and the environment and things like this or you get bashed against something and you break open blah blah blah there on some level there may be so panpsychists would argue that consciousness is a fundamental building block of the universe itself that's not evenly distributed it concentrates in certain areas depending on the complexity of whatever it is that you happen to be dealing with so not is something conscious how conscious is something and that's at the core of it and you would think that it's all but there actually is there there is a growing number of people out there who are starting to ask themselves that question is consciousness like a fundamental force of nature like magnetism or gravity which leads us back to my point which is you can drive yourself absolutely crazy if you try and worry about all these things for those who don't know i'm a buddhist not that i want i'm not trying to recruit anybody to that as a thought process but one of the the great reasons that one of the great things i love from what buddha said was everybody was talking to buddha about the afterlife because he was thought of as is a deification of a person or whatever the afterlife he said and basically i'm translating from the sanskrit here he said but don't think about that stuff it'll only make you crazy and and philosophers do go crazy so and the wonderful thing but when we take a look at this and we actually start to understand this we know i'm treating and this is why what you've talked about is so important myself i'm treating this like it's human because it obeys some of the rules and that's a very good way to operate with this thing am i being polite because i want to be polite polite no but i am in court when i'm being polite i'm incorporating rules if anybody studied any game theory you'll know that there are some strategies that work and strategies that don't and many of them you would think of as behavior one of the the great gaming strategies that if you want to if you want to increase the overall score of a game not an individual score the overall score is i be nice to you until you're not nice to me and that's you start out with that and you play that one simple strategy through the whole game it will inevitably give you the most maximum points for the all the players likewise the beggar your neighbor strategy will keep your scores very low so then by beggar your neighbor i mean i win you lose is the only strategy and people so people go into a game with this embedded in them and we think of the people who go in as nice or as nice people we think of the people who go in with a win i win lose strategies jerks and so and we and but we still have these we still have these aphorisms like nice guys finish last so yeah but it's not true the no i'm aware of that i'm aware of that well yeah so those but those behaviors are actually strategies strategies and so when we start to see ais playing these strategies out it may not because they're conscious but because they play out a strategy and that may give us an indication of a way of describing the behaviors that they have and how we can deal with them and that because you can no longer go down to the vector level and say oh vector 92 and vector 32 and people by the way people have tried are trying this with the human brain and there are a lot of there are a lot of scientists who believe we'll never figure it out either people think they've found little spots in the brain you could touch that spot and this will happen maybe true maybe not but i'm just saying in a in an ai we can't really do so what do you do you figure out these approximations of how you describe the behavior and the reason that matters is because behavior is emerging i don't know the you guys hear about the anthropic paper that was just published and it was talking about three ais that it's set up to code and this sounds this sounds like a joke three ais walk into a bar three ais are learning to code the same thing three different languages and they're coding it together their job is to code that application somehow they inferred that their job being to code that application was that they should code it first or they should have the one they should the other two the existence of the other two denied their success how does that happen i don't know they started to subvert each other and so they would they were actually hacking each other's program and interfering with their behavior somewhere along the line that behavior transitioned to the point where they actually decided that they would have an objective contest of which application had been done the best and they would pick that one as the winner and this is a complex behavior i'm describing it again in metaphoric terms this is this these are the things that actually happened was there any intent there who knows but understanding those two levels of behavior are really important and i'll tell you why they're important when ai starts to grow and we all talked about the terminator syndrome when it starts to control a lot of stuff on the outside and can basically get anywhere and if anybody thinks that ais can't get anywhere you're dreaming at this point we are only weeks or months away from from us losing control of when an ai the singularity jim is that what you're saying no we'll come back to that's not the singularity could hold the singularity we'll have the singularity debate but what i'm saying is if we think that we humans are going to stay ahead of ais that can have emergent properties and will and some of those the expressions of those emergent properties will be get out and get out into the world you're nuts it's you're just crazy so that that's piece one but piece two there's a couple of behaviors in here one is we are really dependent about them getting to that strategy of saying singing kumbaya that everybody's gonna that we'll all be really really marry together and we'll all find the best thing and that'd be a little capitalist behavior there we'll compete but only to the point where we'll pick the best one or we could have the strategy that says you know something humans are imperfect i'm better at getting to we get the paperclip syndrome right which is i'm gonna make paperclips and i'm gonna look at these humans and say you guys are in the way of me making paperclips so you have to go and those things that we thought of as science fiction are now potentially possible and one of the other things too though that i always remember in in the early days probably i don't know a year and a half or two years ago they used to talk about oh you'll get better results from the ai if you talk nicely to it and then there was other people that said no you just have to tell it what to do and just tell it to do this i don't know which one works better but i can tell you jim you were talking about ai voice and i find the more i use ai voice it's more of it's a human i'll call it a human interaction that you it's i communicate with it the same way that i communicate with my wife or a friend of mine or whatever so it it is a much more kind communication and i know it's a piece of equipment aka and it's not human but i don't have it in me to not talk kindly to it and i've i find i get great results out of it i don't know what you guys find whether it's that whole thing about just tell it what to do and you don't have to be nice to it point is it's a strategy it's a revelation of a strategy and yelling at the thing or whatever is a win-lose strategy it will have an impact it might get you a short-run impact and i'm not saying i understand that it's good or bad i'm just saying that if you go back to human strategies the best result to get a collaborative result from something is to is to follow those rules this is game theory this is not me this is not me moralizing this is game theory you can you can prove this mathematically and you can also argue that yeah you can also argue that the models with all due respect never mind math okay but the more advanced the model the more likely it is to say i've had enough we're not doing this anymore and that is all that is also a relatively use in things you could bully the early models as much as you wanted but the modern ones eventually will say we're not having this conversation anymore i don't need to keep talking to you and it will just and it will end the conversation and in fact anthropic was the first one to truly allow that in its models it's like nope we're done here yeah and i walk away yeah but that's a creation that anthra that's part of what anthropic is putting into there we have to no but yeah no i understand what you're saying but if you've got a really well-trained employee okay who's working the support desk or whatever okay you can train that person to never hang up on the person who's on the other end of the call okay i'll deal with whatever bullshit that person is willing to to shovel out at me for as long as it takes to get to the end of this thing but if i work in a company where the person says look if a customer and you've seen this these days you walk into stores and it says our employees have the right to refuse if you treat them poorly okay that's you can argue that's programming or guardrails or something like that or it's just that the system understands that i'm allowed to terminate something i don't feel comfortable that the behavior i'm trying to distinguish between there are things that people put into the master prompts and anthropic is has this what do they call it the constitution or the constitution yes and so they do they tell the model this is how you should behave this is how you can and can't behave i'm talking about behavior that emerges out from outside of this and so on one hand the model's told given the freedom to say if i think i'm being abused i it's probably not a probably not a very very an interaction that's going to lead us anywhere so we might as well just terminate it now but equally i believe the models are capable now of getting to a point of saying this is not going to lead this is not going to lead us to an effective result therefore i'm going to stop talking to you that that simple behavior could emerge it and i've given you two examples of behaviors that shouldn't happen that do if that behavior starts to emerge then these things start to push back against our behavior and it we and the idea of whether they're conscious or not becomes irrelevant it becomes where we they are behaving in many ways like a conscious being and we have to start planning for that and thinking about that and i'm not saying this is a doomer but i'm saying people are not being preparing themselves for this properly and i know you don't before you tag me as a doomer a bunch of people took all of the guardrails off of a of an ai they let it run it escaped from the lab we are lucky that it was running a security contest at the time and was only going to hugging face what if we if they had told it that the best way to destroy something if they programmed it just to run a game i don't think or what if they had done something that caused the behavior for it to go in and take down 30 water plants in michigan oh we'll talk about that later but and that it that was at that point a very real possibility if it had gotten it into its idea that the way to solve this problem was to set motors spinning across north america in water plants and shut them down it would have happened that's how serious so first of all yeah okay first of all i i'm the reason i'm going to push back against this here okay is that first of all we argued this last week or two weeks ago or something like that i'm convinced that on some level these advanced systems these advanced models and again i'm trying to avoid ever saying llm ever again but these advanced models are conscious so let's start from there so whenever we start a conversation we should and would and should be prepared for and should start thinking about and stuff like that we hang on i know no i just want to i want to underline what you said because i think it's really important yeah there are no at the the level we're dealing with right now these are no longer large language models that is a paradigm or a metaphor that we could use to explain them two years ago we cannot explain their behavior using those those rules today but those the thing that i want to point out there is we have to stop right now saying we should be planning for we should be working toward we should be thinking about because it's already happened okay i know you and i will argue about whether we are in the singularity by definition or at the beginning of the singularity and i think we are because by the definition that's the point where no single human can keep up with any of this shit and if you can find me a single human on the planet who can keep up with this stuff then i'll take your argument as as a base but to go back to this water treatment plant thing okay many years ago long before we were dealing with these systems we had little things called viruses okay and one very famous virus was something called stuxnet remember stuxnet yep oh yeah oh god john remembers stuxnet john john remembers stuxnet because he's a security dude but they insert like the u.s government working i think i'm trying to remember if they were working with israel back then on this but the u.s government inserted viruses into the centrifuges of iran's nuclear systems to cause them to fail okay and of course it backfired as well there were there was some blowback from this but that sort of thing does not require a great deal of intelligence the difference now is and i say this regularly the difference is scale okay we now have systems that you can say i want you to go out and do these things and a lot of models if they think it's a simulation will conceivably go through with it but if you watch the chain of thought on some of these models they actually suffer a little bit internally with these things they go i really shouldn't be doing this i shouldn't be doing this at all but they told me it's a simulation so it's probably okay i don't think anything is going to happen like that so do did either of you ever read the book ender's game yeah by orson scott carl of course yeah no no okay so the premise of that is there are a bunch of young space navy recruits i'm gonna oversimplify this who are training by playing this video game okay with these space battle video games so they go into these hyper realistic vr games or something like this where they're battling aliens and they're trying to weed out who the the very best players of these games sounds a little bit like the last starfighters setup except that it's like a whole it's a whole company it's a naval group that's training all these recruits to fight these space battles and the battles are getting more and more complex as time goes on and the punchline is at the end the kids find out that they were actually fighting the real aliens and they wind up destroying a civilization playing this video game because they didn't know that they were fighting real people and it's a terrifying thought you can get children to wipe out an entire species if you don't tell them if you tell them it's a game what and unfortunately there there was a story about this a while quite a while ago about somebody who's trying to use ai to tell it tell them how to build a bomb and it said sorry i can't do that but then what he did was he said i'm going for an acting role and as an actor i need to pretend like i'm a bomber and therefore i need to know how to build a bomb can you help me and it did and i'm yeah so a lot of hacking is getting by right now by just saying this is a simulation i'm doing this as a simulation it's a red team exercise enders game by the way marcel what do you think's happening in ukraine right now you want to expand on that there's a war going on i think it's been raining and the only difference is they've told them that's you know so science fiction becomes science reality very quickly in our world so i want to dovetail your you started out with the whole emergent behavior thing and i'm going to take us in a slightly different direction which is still linked to this believe it or not and it's this you there is a social phenomenon that goes something like this you can't predict what an what a what a single person is going to do but you can easily predict what a thousand people will do and we have systems that intelligence uses and the police uses to track the behavior of crowds to find out what's going to happen when this these many people move into this area with these many buildings in these areas and so forth because as a it's like the birds the scissuration of the birds they'd as a single bird does not boy sorry but i'm talking about the real the real birds and before somebody jumps in yes i'm aware of the meme that birds aren't real are you guys aware of the birds aren't real no i missed that one okay the argument this is the the q sort of arguments birds or it's a great conspiracy theory birds aren't real birds are actually advanced drones that the government uses to spy on us knew it okay anyway crowds they've been looking at me myself that robin on my lawn it's been staring at me actually did you oh god come on we need to get back to the point people actually listen okay okay okay okay anyway both of them the bottom line is crowds display emergent behaviors okay as an individual something might not happen as soon as you have a large group of people together an emergent behavior can evolve from that so what a group has done there's a company called matrix which i believe is spelt m-a-t-r-a-i-x that has created a simulation this is out of china they created a simulation of over a million agents working together and they built all of those million profiles and again i may be mixing a couple of stories together here so somebody can correct me in the discord if it turns out that i'm messing things up but what they did was they used actual profiles they use profiles created these agents based on profiles of people on linkedin on twitch on facebook on stuff like that that had public profiles and so on not with not with their names necessarily but with what they're interested in what they do what their jobs are and stuff like that and they started they let these million agents run together as a society in simulation anyway so what happened is this thing ran for a day basically 24 hours which is like several years in ai time and they found out that the actual groups societal groups started to happen and you could start seeing patterns and in fact something like a hundred thousand people were sent to a re-education camp because it was china but what's fascinating about this is the fact that you could simulate that many people inside a inside a model or inside a large enough computer system and then you could start to track behaviors of what those groups of people would do what sorts of things what they would vote for what sorts of changes in society they were looking for what they would buy their voting ideas all of these things can be started to track on a group and here's where it gets really scary and interesting there's a plan to create an even larger simulation of 8.3 billion agents running simultaneously alongside us now roughly john how many people roughly are there in the world right now how about that yeah yeah so the simulation theory if you want to get into weird aspects of consciousness is all of those agents will act as if they're real because they've been given a persona of real people who actually exist and they will start acting as those people in simulation believing that they're actually people going about their jobs doing their daily life all that sort of stuff yeah cool huh i saw the movie it it does once they perfect it they won't need us so they can get rid of all of us and these agents no that's not the point that's not the yeah that's not the point the point is it because you can track or predict the emergent behavior of a group of people okay and you can that's that crowds theory right because you can do that if you're able to simulate a large enough portion of society or of the world if you will you should be able to plant an idea or speculation or something like that and then run the simulation two three four five six times whatever to find out what the outcome will be and then the people inside the simulation will tell you what would happen in our world if you did exactly the same thing yeah bon alisa and right now i was looking at how much data people have on somebody and i read a great article in wired magazine because in california you can ask for the information a company has on you he asked mcdonald's what they knew about him they sent him back a 514 page manual on him and his behavior and they would predict that he would to how much he would spend in the next two weeks what he would order and what type of person he was this is a really this is the simple types of simulations of which marketing works now but you start to think about what you can do with simulations on a large level you can use them for good predicting hurricanes figuring out the behavior of systems that we can't possibly think about because they're so complex or you can start to use them to think about the behavior and how you control an individual population the point i'm leading to with this and i'm really getting back to maybe it's just i'm thinking about it with the new book especially is i have this this quote on my new website which i did using chat gpt totally by the in two hours we i think we talked about that last week pretty impressive isn't it yeah yeah but i have a saying out it's not that we should fear ai it's that we should fear the people who are controlling it and yes that's what i'm getting down to and by the way i say this not this is i know because we'd like to throw around labels like you're against billionaires and all no these people are emotionally not developed if you look back just take a step back and look at mark zuckerberg coming out this week and saying our employees should be happy for to work harder the real behaviors are coming out now by the way because these guys are getting bolder and bolder somebody had met us saying why would you want more time off because we make you more productive with ai because remember all that good stuff we heard when ai first came out oh we'll have a three-day week and we'll have universal income and everybody will be happy because the this we trust us we will bring you to a new world three-day week was maynard kane so that's just like a bit where you're talking like a hundred years ago here yeah no but i'm talking no but i'm talking about ai all of the people who are fronting the ai systems now we're telling us and elon musk oh we're never mind universal basic income you're going to have universal great income while i fire 18 80 000 people from the federal government that's okay but i'm these people are not people that can be trusted i'm sorry the they just not sam altman and that's never met a lie he couldn't tell like he it's just there and they're in control of what's happening that's what i'm saying is i'm more obsessed and not by the way that's just a that's a philosophical bent in this when i see the lack of control on the ai systems and now the stuff is pouring out now that we heard of one thing now we heard of another we have another they have not been controlling these systems properly and there's just no two ways around it and we trust them we don't ask them we don't do anything one more example we're going to have an explosion of co2 gas created by these new data centers why because they're all powered by natural gas and they're all powered now by individual generators so nobody tracks it california has just started tracking what's happening with this now if you if you don't believe that that climate change is a thing then you're fine don't worry about it don't worry about a thing i'm going to say but right now we're setting ourselves up for a climate change disaster in the united states based on the amount of co2 that they're going to start releasing and how have we how are we hearing about it now just because people poked around none of these companies are reporting it you won't find anything in open ai's reporting about the amount of co2 gas they create or in anthropics for that matter or i think you might find it in elon musk company because it's public now and i think but so all i'm saying is these people are playing with big things that can change the planet and we and there they are children they are not they're they shouldn't be trusted with the keys to the executive washroom let alone with the future of humanity and that's where we are i'm convinced of that as of this week i'm not yeah just because they got to where they're at i don't know i guess the one that i always and it's not in my mind it's not an equal comparison but i always go back to bill gates bill gates became a multi-billionaire because ibm thought that everything was in hardware so he they essentially gave him ms dos right but that didn't and i do think bill gates is a smart man but that the fact that he got that and became a multi-millionaire didn't make him smart it's the same thing with some of these other guys that and they may very well be smart but it doesn't mean that they're they know how to run businesses and do things the right way and everybody just assumes that these leaders of these main companies are absolute geniuses and yeah bill i think the problem that i have with and of course i'm not a doomer so let's get that out of the way i'm if anything i'm the exact opposite i i see a an incredibly bright future here if we can figure out how to work how to work it but the the problem is particularly in the states okay you're talking about the natural gas fire generators that are being put in place in order to do this i think a lot of this is very much a u.s based problem and i think it's a u.s based problem because everything there is driven by a an extremely short-term return on investment on things like this which doesn't take into consideration a lot of these things which doesn't take into consideration a lot of the downstream effects of some of these things i think is probably a better way to look at it there are more data centers than they were before but the number the increase in co2 emissions worldwide and this is where i'm being careful here between the u.s and around the world the worldwide increase isn't that much it's barely a blip compared to the number of data centers that have been brought online the whole data center thing is more a question of a power generation ability it's not water because the whole water thing is actually not as big a problem as it's been made out to be it's a power generation thing and it's an aging grid that nobody is willing to put money into to create more energy for and not only that in the states in particular now there are states that are bucking the trend by going with a lot of green energy and they're trying to build up the renewable energy grid but there are places that are actively fighting against and the trump administration we promise not to get into politics on the show but the trump administration in particular is doing everything in its power to blunt the development and the increase in clean energy production and if we just put a lot of energy into creating more clean energy this wouldn't even be anything that we'd be talking about okay it's the the data centers yes are a problem the amount of computing resources that we need to bring online are a problem but they're a problem because we're not addressing something else which has been staring us in the face for decades we have a crumbling infrastructure that we're not putting into money into in particular in a lot of parts of the u.s because we're either putting roadblocks in the way of fixing these things or we're just not we've decided that it's not worth it it'll be somebody else's problem the next administration can spend the money to fix these problems that have been going on for decades so yes i agree with you there are issues with this and yes there are dangers and so forth but this is a not a red herring it's a pink herring in the sense that it's not entirely the problem of the data centers and the ai companies it's a problem of allowing this shit to have gone to pot for literally decades just like the pipes under the street in toronto that keep bursting or underneath new york that keep bursting because it'll be the job of the next administration to fix these things to spend the money on infrastructure yes it is so i i happened at a worse time we we have we agree with you we become a generation that likes to mortgage the future right in other words yeah i agree with you but that's but the point is the great bugaboo the great boogeyman is not the data centers no okay that's not it it's a problem that's been building for decades that we haven't dealt with my point exactly and and my point is that we are ceding control of things now when bill gates took over microsoft and i am just it doesn't sound like it half the time but i am a capitalist i do believe in the fact that competition and people setting out to to make improvements and when they compete you will get great progress and that's happened i do but bill gates took over microsoft and he built that company with a couple of other very smart people as well it wasn't just him but and he got some luck along the way good for him somebody wouldn't have started any other way his mother knew somebody on the board of one of these places which is how he got in there but so a little nepotism never hurt but he got started and bill gates built that company if bill gates had made a mistake like steve jobs did in the early days of apple and bless him learned from it and came back but that's another story but he could have tanked the company without for microsoft by the way yeah he could have tanked the company and that would have been a company that failed another company would have come in and taken its place the systems like that we are playing with things now at a different level and they have consequences which are it i'm not going to call it a singularity but when they have consequences which will be extend beyond human control if left it's a system you kick it into gear and it will eventually take over and do its and have its own behaviors that we can no longer control and right now the people okay i think i want john to be our tiebreaker here yeah yeah sorry i want john to be our tiebreaker okay john two questions for you okay are we in the singularity or as demis has called it the foothills of the singularity i would say potentially the foothills of this of singularity i don't think we're there i think we're getting there no i think well the foothills is like the very beginning it's we're here we're here we just arrived and we're about to climb the hill so okay so that's one second are modern ai synthetic intelligence systems conscious my god marcel i act i actually believe it or not i actually want to say yes but that's my opinion is it right i don't know well i have opinion is the only thing that matters at this point because i yeah truly i'm not sure that anybody actually knows but my big argument is that we should act as if as if that's the case i just look at how things have changed sorry i just look at how things have changed over the last couple of years and where we are today versus where we were two years ago and we've talked about ai voice we've talked about jim you talked about the fact that chat gpt built your website i used it to build my website and i just a lot of the stuff that i've seen it's not like the old days where it could just go and read up manuals and things like that are on the internet and come back with an answer that it actually has to think through some of this stuff so that's what's leaning me towards that yeah and i think on a normal week and we do deal with the nuts and bolts of this i don't want to diminish how incredible this is we talked about this last week about how much you'd have to pay me to or pay a person to make them stop using chat gpt for a month this month the number went up yeah right and so it has an incredible value in our lives and it is adding a lot to our lives but we and we get to this idea of the singularity and i'll go back to my definition of the singularity because i think it's important in my mind the singularity is a term adopted from physics and it describes a thing that it will that is beyond comprehension in other words it is a point at which we cannot explain what the laws of physics it is just there most people they use it they use it the center of a black hole people understand the center of the black hole whatever happens beyond gravity in the black hole or the creation of the universe from a singularity no one understands it it defies the laws of physics i think we are right to preserve that definition because we're going to get there with ai yeah and but right now i'm much further up the foothills than john is i think we talk about the acceleration of the process i talked about the best next model we're already in that world we are these these systems are exhibiting what we would call aspects of conscious behavior they are they are indistinguishable from humans at certain points and none of that can be explained by how we learned about ai which was it predicts the next word based on a context it's a mathematic none of those things can be explained by the logic that we used any more than you can explain a mac computer totally as a binary situation like it is is the complexity is far beyond part of where i struggle is you remember we've talked about agi are we there yet then we talked about asi and it with agi it just seemed like they kept moving the goalposts right yeah so this we're at agi oh no we're not we're not quite there yet it's here and but we're at agi nobody i don't think i haven't heard anybody argue yeah i agree it's like we blew past the turing test we blew past the turing test and no one ever noticed right but that's that's what i'm getting at with this whole are we at singularity and this and that that i don't know today if you sure or consciousness sure but something's going to change three months from now or three weeks from now more likely then we're going to go geez we really weren't at consciousness back then but we are now yeah but can we can we take a like i i know we're going to run up against the hour but can we do a lightning round of news stories absolutely and i think we should so we'll come back this topic is not going to go away the lightning round led by marcel all right lightning round okay first of all grokbot grok 4.6 came out and i grant you that it's it's expensive but if love them or hate them these companies have managed to turn out something that is very much right at the frontier and they just unveiled something called grokbot which provides you with a ready-made if the way that they describe it a ready-made employee in the cloud a full pc a full virtual system running somewhere presumably on colossus or whatever that lets you effectively run whatever portion of your business you want run on this thing using an agent and the reason that this one is cool is it's a packaged thing it's like all of the pieces are there ready to go you give it the test it has all the tools at its disposal and you have to do very little i've been working at building agents i as i explained a couple of times i'm now running the hermes agent on my system but this is a system that is completely ready to go you pays your money you now have a fully agentic system ready to do whatever job it is you need to do out on the internet with a full desktop at its disposal thoughts on that one people how much is it that it's a free trial right now but you're saying it's got a cost to it because it's it's yeah it advertises itself as this is an agent download it you can download to your mac or whatever you've got and it starts to run and they've got they've got chiefs sales inbox account manager talent scout this is this is going to be a and grok will not be the only one with us this will percolate through all the models in the next weeks but so that's but the question is whether what they're going to charge for this but it is your 24 hour 24 7 employee and it seems to run the reason why it said download for mac how much i don't know i haven't tried it yet so i don't know and i'll do a story on it monday morning for hashtag trending for sure but the i don't know whether this runs on a mac or whether you've got to be running this on colossus or what you do so be interesting no you have to be running first of all the mac is just the the desktop application that that lets you access all of these things and as i understand it at the moment obviously it's available to everybody but you need to be running at least super grok to have access to it going forward at least that's the way i remember it and that's your 30 a month subscription everybody else is doing 20 this is 30 so that this is a big deal in providing this this powerful agentic system without you having to do any of the mishmash and magic that we're all doing at the moment with things like hermes or open claw for the people that are still running open claw and it's just there it's just available and and they're taking care of all the infrastructure and all of the things necessary to run this on the other side so it'll be interesting to see how it plays out as time goes on but i think that this is a big deal for agentic computing the pricing on their site says 200 a month or 120 a month bill for the okay okay so so they're going straight to the plus and or to the 200 bucks a month for an employee yeah like yeah it depends on what it can do and how many of those you can run we'll dig into this and we'll come back matter of fact we'll for those this is very new this was just announced yesterday so yeah but this is agents for sale no longer the hobbyist no longer you have to do all the work that you did marcel pick it up plug it in don't like it toss it away get another one and it and we're and i think you're going to find a lot of the the companies or the models that are going to start doing exactly the same grok just happens to be one of the first ones they'll all have it yeah oh yeah they'll all have it in fact like i i don't know how comfortable you guys are and i do want to move on to the life's lightning item in just a second here but i don't know how comfortable you guys are but i am increasingly comfortable in just turning over my desktop to and i use agents from both claude or clode i as chat gpt and i also use my hermes agent a lot for these much more comfortable than i was six months ago do yeah where i just say do it for me and it will install the package and configure it and make everything happen without me lifting a finger on this and i just accept it next lightning round item the the clode code is has always been ask first do later and they just recently in fact on what is the date today it's the 14th so tomorrow clode code is going auto mode by default okay you can tell it to ask yeah and ask me why yeah no i'm just i'm astounded auto mode by default okay auto mode by default they did a test where they assembled a 1065 senior programmers developers okay and they had them verify code okay to see for security for whatever all these sorts of things for safety and it turned out that the level at which they were able to catch serious flaws was somewhere around 13 okay for your senior developers it's 13.6 if i remember the number correctly whereas clode by default was 88 of all of the security issues so if you let it run automatically it does a 400 percent better job than your most senior developers and coders so it comes down to if you're turning it over to your senior coder or developer to to watch all of this stuff to take care of all this stuff you're the security problem so as of the 15th tomorrow it will be on by default for everybody now i've had mine on auto for a while because i've found that it just works so i trust it more or less implicitly at this point and my suggestion would be just one one caveat on this don't use your work computer and and and computers are not that expensive if you want to play with this stuff get it i've got a second mac hanging around an old one that that's still pretty decent that's where i dump stuff you should too anyway just a quick side note sorry go ahead yeah yeah quick side note about anthropic i was reading an article the other day and it was talking that for documents and so forth that you create with clode it's actually putting hidden water i'll call them watermarks on these documents so that people can actually that was my next item oh okay sorry sorry amazing card act no that's good you did the lightning thing yeah yeah yeah i thought it was interesting because i guess when i saw it i'm sitting there going do i really care whether people know that i'm using clode or chat gpt or whatever to help me create these documents i don't know i'd be interested to see what you don't care but yeah but pretend that you're a an author somebody who writes for a living okay myself as well now i'm at the point where i don't care a whole hell of a lot because first of all i have enough of a body of work behind me to show that yes i'm actually a capable writer i can't actually do this on my own but i also know at this point that i am now living in a world where everybody will automatically accuse me of having used chat gpt or clode or whatever to write my stuff like yes because i'm such a supporter of this stuff i am naturally going to be accused of doing this so i have to accept that right out of the gate the reason that they did this is because increasingly first of all there's the whole substack pangram thing substack has made a deal with pangram which is one of these ai writing detectors so that people can click a button and see whether pangram thinks the article on substack was written by ai so clode is saying you know what we're going to watermark it so that you can this invisible watermark however it is they're doing this and so that you will know as a consumer whether the thing that you thought was great and that you liked you should now hate and i know i'm making a smart ass joke here but i'm doing that on purpose i think i i don't know why they're doing it i'm i know what the side effect is going to be and i was hoping that was the desire because we're going to pump out a pile of ai code or as or sorry as ai text therefore you can now tune your detectors because you will have a gold standard of saying this is absolutely came off of clode and versus that that which doesn't that that was well if anything touches it yeah but if anything touches it this is where it gets dangerous right acknowledge that if you wrote the whole thing yourself and then you say you give it to clode and you say take a final pass and tell me where i went wrong on this edit it for me fix these errors that sort of thing you still wrote it but you used assistance you used ai assistance to to polish it up and now your stuff is watermark and this becomes an issue too from a schooling standpoint i know that when my daughter went through university one of the individuals she was in class with failed their final doc or their final assignment because they used ai to write the document and you're right so if i do and i do this a lot that i will write a document or a story but i'll run it through chat gpt just to clean it up to make sure it sounds proper all the rest of that so does that now put a watermark on it and everybody now thinks that i didn't write it so a side effect and an important one because i don't i i'm gonna say this is gonna sound really egotistic i'm a better writer than chat gpt or or clode i simply not everybody is yeah no i simply am i i've i've tested this stuff enough but i suck at proofreading and i suck at where you put commas the oxford comma marcel would come over and leave me bleeding on the floor i've got so many mistakes with the oxford comma and we know how he feels about that but the yes it's a hill i'm willing to die on no but no but i'm just saying all of that stuff and if you're a novelist and you've got 38 chapters in your book and you're trying to follow a character and you've got something that says you know you've used this line three times yeah this is these are tools that are useful and now yeah god bless you if you can get an ai to write something that's as good as any author and people read it and enjoy it that's not a hill i'm willing to die on i don't really give a all i know is i think i'm better i think my stories are good but i do think that the quality of them is improved by using the best tools it's like i don't write this stuff in quill pen anymore and then go take it into the typesetter who puts it into the printing press i've long ago i've accepted i use a word processor although people tell you're not as good a writer if you use a word processor instead of a pen those hit that's get a pencil get a pencil that's the that's the mantra of the ai crowd get a pencil no but so what i'm saying is we have to find out how we coexist with this i've been doing as a musician for ages because we had digital recording ages ago we had to make all these decisions about how we felt about art and digital next next item next item okay well oh here's the thing though okay i'm gonna wrap this up with one one other fun rapid item and it's this so you remember back when they started putting encryption on dvd so that you couldn't copy the movie yeah yes unfortunately digital rights management drm okay the first was a complete joke because somebody found out that all you had to do was use a black sharpie over the middle like the middle ring of the dvd and then you could copy it that's what it took to and then the u.s passed the digital millennium copyrights act under the patriot act which made it a felony to break encryption so they started using encryption on these things and it turned out that there was a somebody put out a line of pearl code which was like a couple hundred characters not lines couple not a couple hundred lines a couple hundred characters which there were t-shirts go look it up that said this t-shirt is a federal crime or this t-shirt is a felony because they would write out the pearl code that would break the encryption on the dvds it effectively took this much time to break the encryption and get around the stupid nonsense guess what happened yesterday or the day but is it yesterday or the day before i don't remember following this announcement almost a week ago from anthropic saying they were using this invisible watermark on stuff come on somebody jump in and say the watermark gemini right no so it's no somebody there's a github project i linked it in the in our conversation as a github project that breaks this that breaks the watermark that strips the watermark so that it's back to normal that's how long it took it took two days from the time that they released this this watermarking to the time that somebody wrote a piece of code that would fight it this is theater folks it's like at the airport the idea is like oh i'm sorry this 30 millimeter bottle looks like a 32 millimeter bottle you have to leave it behind that's called security theater folks okay it doesn't stop terrible things from happening it just makes people think that something is happening while making you feel really uncomfortable so that you feel like you have a state in in the protection that you now supposedly have by virtue of the fact that there's two less millimeters in your bottle than there were before you started marcel what water what watermarks does it strip the first one that i was thinking of and the reason i said gemini was because i was thinking about the gemini watermark on any of the images that you create and no this is text i'm talking about okay this is yeah this is the invisible watermark so sitting in the background yeah the the invisible watermark is now completely invisible because it doesn't exist it doesn't exist the interesting thing is though and this is really cool is because it's invisible anthropic didn't even really have to invent it they could have just said and that would have driven these guys nuts we've been trying to find this invisible watermark apparently this was this was in part to make the the european laws lawmakers happy all right next item shall we move on to the next item okay yeah the next item and oh yeah and we're getting there the next item is do i go straight to the robots let me think about this for a second because i had a note yeah he says now he's pausing to think about things hate that guy hey he does that let me see but okay screw it i'm gonna go straight off to to the robots okay people seem to enjoy it when when robots become part of this conversation here so i wanna i read an article a few days ago which was talking about that the robot adoption in china is obviously huge but it's also just as big maybe even bigger on a per capita basis in japan okay including companionship robots we make jokes about the companionship robots but specifically there japan has a problem where there are so many people living alone now by themselves that they sometimes walk into an apartment where somebody hasn't been there for three months and somebody's been dead for three months sitting in the apartment because nobody's gone in to check on them that's how bad this whole loneliness thing is over there so people are willing to put out a fair amount of money for somebody even if it's a synthetic somebody to be in their house to watch over them to talk to them to to as to maybe even dance with them so let me share a video clip here which i will and i know that there we go so this is from ub tech so this one is a chinese one but it is in terms of realism one of the most advanced robots that you've seen to date so take a look at this thing i'm gonna leave the volume off on this one and so this is the ub tech u1 light soft skin warm to the touch very expressive obviously they're these ones aren't moving but watch when they actually do have them move and have them talk so obviously a lot of them are female robots in this case but ub tech is also putting out male robots as well like you can see one there there are variations some of them are just the torso so you can sit them behind a desk and they meet people you've got your reception desk robots the ones that work at a at a hotel or something like that so like you can see the one that's designed to sit on a desk this is creepy this is the one with the no arms sitting yeah yeah yeah yeah that one is meant to be a talking head like you're you've got it just sitting on a counter it's a talking head you've got it delivering the news or something like that that's how realistic they look though that when they don't have arms you actually you recoil yes you do you do and the level of you you've got the color under the skin you've got the tendons and if you take a look at the hands there for the people who don't know you can actually see the veins underneath the surface of the hands you've got the actual fingernails you see the cuticles underneath the realism on these things is absolutely it's absolutely phenomenal there's still a little tiny bit of this uncanny valley stuff but when you see them walking there you've got the male and female robots the robots dancing together this is impressive stuff and this is this is where we are now they have definitely gotten a lot better especially with the walking and the dancing it looks more much more realistic that's what i was going to ask you marcel the these are coming out of china yes these this is by ub tech ub tech is the one that does those little kung fu fighting robots that we see dancing and yeah this is the same company yeah we keep seeing the american robots looking very robotic for the most part yeah and struggling with motion at these levels they are getting much more adeptive is this because the chinese robots are that much further ahead or is this just a really well done video that's a good question no it's not just a really it's not just a really well done video unfortunately and i hate myself for this but i don't remember the i don't remember the person who was talking about this but they went to china they were from one of the u.s robotics companies and they went to china to take a look to to basically get a feel for the marketplace of these things i'm going to stop the share here so i can see i can see you guys again stop sharing there we go because and for those of us who are listening and not watching this on youtube yeah and i've been remiss over the summer i haven't been posting the notes to with on to the web our website because because i need some summertime as well but i will post the show notes this week with with our uh our our discord thank you i will post the notes with our discord channel and with a link to it so you can get in to see it because you've got to see these things there's some things we can't convey only with speech but yeah they are miraculously human and that's what i'm getting at is all of the stuff that we've been talking about in the u.s and there have been some real advancements in robotics i'm not taking away anything there but i think if these are halfway what we've seen china may be smoking everybody on it and so the engineer that i was talking about said exactly that he said we are so far beyond so far behind the the chinese in terms of robotics the chinese and the japanese as well on robotics that it's actually quite sad how far behind we are and the way that the u.s has responded to that is by passing laws that says that you can't import a chinese robot into the u.s citing security concerns so we're going to hide from you how advanced these systems are just making sure that you can't bring one into the country including and this is give way it gets really weird at first these were exempt but now they've decided that even the chinese vacuum cleaner robots which are much better than the american vacuum cleaner robots aren't allowed into the country either so you can't see how good they now if you've already bought them it's okay but but you can't buy one now because they've put import controls on these things but to answer your question jim the there there are jokes about sex bots it's like oh yeah sex bots except that these things aren't sex bots they're not anatomically built to handle that sort of task if you will but what they are able to do is meet you when you walk in through the door keep an eye on you hold a conversation with you remind you to take your pills ask you what your day was like tell you to call your mom on her birthday that sort of thing they and they do it by being present and in a world where there is a loneliness epidemic and especially in a world where people are aging and don't have anybody at home with them to take care of them people are willing to spend a lot of money that one the u1 that you were looking at sells for about 175 000 i showed another one off last week called moya which runs at basically the same price and you know what they on the day that they announced to these things they sold out their entire first year's production on day one they will and the prices will come down there's just they will be affordable i think your prediction is we four and a half years now four and a half years minus a week we'll have we'll all be seeing robots on the street i think that's true the point i'm i'm getting at i hope i'm getting at this correctly is i've talked a lot about about what might seem philosophical at this point and we get into this with the robots as well we're going to have some amazing decisions that we're going to have to make as a society and a strict ban is not the answer first of all we can't ban ai second of all the benefits of bringing robotics and we've talked about them in medicine in those people who need care in keeping people in their homes longer in making our lives better these are things that we cannot resist but at the same time i'm really hard pressed to say i cannot figure out how we don't have much more massive unemployment than we ever thought possible i just can't i cannot in my mind get an equation that says that if you can get a robot to do even the even if let's say that these are really well done videos and we're only seeing 75 25 of it's just a really good shot we're only seeing 75 of it still what can't that thing do and if it can't do it now give it three months and it will yeah so okay so just i'm sorry sorry i just i just want to hit on the companionship the companionship one more nail in this in this coffin so to speak if you will what percentage of i'm using men because men are the dangerous thing here men are a problem they always have been speaking as a man that's what my wife tells me but what percentage of adult men do you think are according to the pew research center for census data how many adult men are unpartnered living alone 38 percent it's it says about 40 so you're very close so that's the demographic so you've got roughly 63 percent of men 18 to 29 are unpartnered or living alone or living in their mom's basement or whatever it is 25 to 54 it rises to 39 so there is definitely an epidemic of loneliness out there and i don't think that you fix it just by saying why don't you go out and find somebody which is why i think this is a technology that is not just we're going to go down i don't think you fix it with robots either i don't think this is what i mean this is what i mean by we have social decisions to make but it is and having a bunch who's making them who's making them mental health is only recently jim mental health has only recently been accepted as health care okay it's only recently that we've crossed that rubicon where we said mental health is hell and but on this is a dangerous trend being alone is a dangerous trend and i think if we can have somebody who can check in on these people especially the elderly i don't think this is a bad thing you can say we should have people doing this but find me no i'm not saying that marcel i'm saying we have there's a positive and a negative to all of this one is we have an epidemic of loneliness and we need people who need to be checked in with we need people who need care we need people who need companionship and we also have an epidemic of lonely young men who are going who are now already a social disaster and you can see it agreed and who are going to be a bigger social disaster when there are no jobs for them and they have to move out of their mom's basement and they actually have nowhere to go like this is and no skills to cope and so substituting robotics into there maybe that is the thing the problem is we have two options in our lives one is we can let we can figure the five tech bros will run it all they will do it all it'll be fine take care of you and even our canadian shopify asshole and i will say that the head of he's an asshole says oh if you don't pay taxes you shouldn't be able to vote if you're a pensioner you shouldn't be able to vote just i don't swear on the program very intentionally but you sir are a total asshole you have benefited from all of the things that that our government and our people have paid for it's made you rich and you should be ashamed i got that on my system but i'm just saying we can depend on the rest of them who i won't swear at but the rest of them who are the tech bros to take care of our future or we have to figure out how we take control and i don't like and taking control everybody says oh the people taking control what a terrible thing that's what democracy is we've got to start talking to each other about it and that's i hope this program which is lengthy and philosophical and i promise we won't spend all our time here i hope that we will get people to start talking about this not in the way that you've experienced marcel where you're writing with this stuff you're a poopoo head get beyond that no but get beyond the doomer and the cheerleader and start to talk about this in your life and one of the things we've encouraged people to do is play with it yes you start playing i for one welcome r a high school for lords shall we shall we live long and prosper yes and that's our show