IoT Coffee Talk: Episode 217 - Tech Betrayal

1 1 [Music] nice you know I think that's the best nice you know I think that's the best recording that we have of you playing recording that we have of you playing recording that we have of you playing it's amazing there you go M yeah hey it's amazing there you go M yeah hey it's amazing there you go M yeah hey everyone welcome to iot Coffee Talk yeah everyone welcome to iot Coffee Talk yeah everyone welcome to iot Coffee Talk yeah so so so um uh remember to like share and do all um uh remember to like share and do all um uh remember to like share and do all that Social Media stuff which is want to that Social Media stuff which is want to that Social Media stuff which is want to tell you about up front so that we can tell you about up front so that we can tell you about up front so that we can get to a gajillion a gajillion get to a gajillion a gajillion get to a gajillion a gajillion subscribers subscribers subscribers there yeah we have we have what I there yeah we have we have what I there yeah we have we have what I think uh is this 7,000 minus a gajillion think uh is this 7,000 minus a gajillion think uh is this 7,000 minus a gajillion yeah yeah to go so plus all the all yeah yeah to go so plus all the all yeah yeah to go so plus all the all it'll take a little time yeah it'll only it'll take a little time yeah it'll only it'll take a little time yeah it'll only take about you know 12 billion years but take about you know 12 billion years but take about you know 12 billion years but that's okay we can weigh yeah so you're finally home man

weigh yeah so you're finally home man you've been traving a lot right you were you've been traving a lot right you were you've been traving a lot right you were spent about a week in Milan which was spent about a week in Milan which was spent about a week in Milan which was awesome yeah and uh good coffee good awesome yeah and uh good coffee good awesome yeah and uh good coffee good gelato and uh all that good stuff and gelato and uh all that good stuff and gelato and uh all that good stuff and meals and you know everyone in Italy meals and you know everyone in Italy meals and you know everyone in Italy they take their food very seriously they take their food very seriously they take their food very seriously that's the one thing it doesn't matter that's the one thing it doesn't matter that's the one thing it doesn't matter where you go oh yeah pop up shop where you go oh yeah pop up shop where you go oh yeah pop up shop whatever everyone's like okay slow down whatever everyone's like okay slow down whatever everyone's like okay slow down this is going to be this is going to be this is going to be this is going to be this is going to be this is going to be good so I I really appreciated that uh good so I I really appreciated that uh good so I I really appreciated that uh everywhere you went the food was real everywhere you went the food was real everywhere you went the food was real good so yeah you know I think I had a good so yeah you know I think I had a good so yeah you know I think I had a roommate uh in college he's from Italy roommate uh in college he's from Italy roommate uh in college he's from Italy and the guy would never eat anything and the guy would never eat anything and the guy would never eat anything when we went out to a restaurant I mean when we went out to a restaurant I mean when we went out to a restaurant I mean he L literally starved just drink he L literally starved just drink he L literally starved just drink whatever water basically U but when we whatever water basically U but when we whatever water basically U but when we get back home he would cook and he would get back home he would cook and he would get back home he would cook and he would go to these you know like Whole Foods go to these you know like Whole Foods go to these you know like Whole Foods equivalent like you know Farmers Market equivalent like you know Farmers Market equivalent like you know Farmers Market type yeah grocery stores right and get type yeah grocery stores right and get type yeah grocery stores right and get his food there he would make everything his food there he would make everything his food there he would make everything himself and I asked him dude why do you himself and I asked him dude why do you himself and I asked him dude why do you go through all this trouble right go through all this trouble right go through all this trouble right because I can't eat I can't eat food in because I can't eat I can't eat food in because I can't eat I can't eat food in America America America oh my gosh that must be like really oh my gosh that must be like really oh my gosh that must be like really really tough but he just couldn't stand really tough but he just couldn't stand really tough but he just couldn't stand how synthetic everything was and POS how synthetic everything was and POS how synthetic everything was and POS it's processed yeah yeah yeah and and it's processed yeah yeah yeah and and it's processed yeah yeah yeah and and you get when I got back from the UK so I you get when I got back from the UK so I you get when I got back from the UK so I was in UK last week right actually over was in UK last week right actually over was in UK last week right actually over the weekend um leis Hamilton yes

the weekend um leis Hamilton yes the weekend um leis Hamilton yes congratulations for congratulations for congratulations for winning the British Grand Prix but um I winning the British Grand Prix but um I winning the British Grand Prix but um I think it's like his eighth win if I'm think it's like his eighth win if I'm think it's like his eighth win if I'm not not not mistaking uh I am not a F1 fan yeah but mistaking uh I am not a F1 fan yeah but mistaking uh I am not a F1 fan yeah but I did go and check it out cool and I I did go and check it out cool and I I did go and check it out cool and I froze my butt off in the um in the froze my butt off in the um in the froze my butt off in the um in the stands yeah it was cold yeah and yes it stands yeah it was cold yeah and yes it stands yeah it was cold yeah and yes it is is is summer summer in the UK it means cold summer summer in the UK it means cold summer summer in the UK it means cold yeah yes yes yes San Francisco too you yeah yes yes yes San Francisco too you yeah yes yes yes San Francisco too you yeah and that's where I am uh today I'm yeah and that's where I am uh today I'm yeah and that's where I am uh today I'm in San Francisco and this is the mosc in San Francisco and this is the mosc in San Francisco and this is the mosc center people starting to trickle in you center people starting to trickle in you center people starting to trickle in you of course today is of course today is of course today is sustainability sustainability sustainability day so I sat in a rather empty keynote day so I sat in a rather empty keynote day so I sat in a rather empty keynote Hall uh it's the CEO Summit thing right Hall uh it's the CEO Summit thing right Hall uh it's the CEO Summit thing right over there I see that a lot of CEOs in over there I see that a lot of CEOs in over there I see that a lot of CEOs in there yeah there are I mean that's where there yeah there are I mean that's where there yeah there are I mean that's where all the CEOs are presenting and of all the CEOs are presenting and of all the CEOs are presenting and of course you know it's course you know it's course you know it's sustainability morning and it's almost sustainability morning and it's almost sustainability morning and it's almost empty and and then who who do they have empty and and then who who do they have empty and and then who who do they have as a guest as a guest as a guest speaker uh speaker uh speaker uh Rick oh Rick Rick oh Rick Rick oh Rick B yeah Al Gore oh good for him like B yeah Al Gore oh good for him like B yeah Al Gore oh good for him like right right well of course it's a tape right right well of course it's a tape right right well of course it's a tape thing right and he's probably he would thing right and he's probably he would thing right and he's probably he would have been glad that he didn't show up have been glad that he didn't show up have been glad that he didn't show up because he wouldn't have been uh because he wouldn't have been uh because he wouldn't have been uh speaking too much of a crowd but yeah

speaking too much of a crowd but yeah speaking too much of a crowd but yeah you got to give credit where credits do you got to give credit where credits do you got to give credit where credits do right regardless of what your politics right regardless of what your politics right regardless of what your politics are I mean the dude was right uh unless are I mean the dude was right uh unless are I mean the dude was right uh unless you're still denying climate the impacts you're still denying climate the impacts you're still denying climate the impacts of you know climate change yeah which I of you know climate change yeah which I of you know climate change yeah which I think is a it's a shrinking fast think is a it's a shrinking fast think is a it's a shrinking fast shrinking minority but um yeah things shrinking minority but um yeah things shrinking minority but um yeah things are getting real and real tough for a are getting real and real tough for a are getting real and real tough for a lot of people around the world because lot of people around the world because lot of people around the world because of um you know extreme weather I mean if of um you know extreme weather I mean if of um you know extreme weather I mean if you want to call it extreme weather call you want to call it extreme weather call you want to call it extreme weather call it extreme weather extreme something it extreme weather extreme something it extreme weather extreme something goofy is happening we're having like goofy is happening we're having like goofy is happening we're having like likeing tropical storms in San Diego likeing tropical storms in San Diego likeing tropical storms in San Diego something's not right that's not right something's not right that's not right something's not right that's not right very wrong by the way I met Al Gore back very wrong by the way I met Al Gore back very wrong by the way I met Al Gore back in in in 1988 uh really he was running for 1988 uh really he was running for 1988 uh really he was running for president if you remember at the time it president if you remember at the time it president if you remember at the time it was like the prim like dick Gart and I was like the prim like dick Gart and I was like the prim like dick Gart and I don't know don't know don't know duckus and I was going to school in Bost duckus and I was going to school in Bost duckus and I was going to school in Bost University I was at Buu and we uh we University I was at Buu and we uh we University I was at Buu and we uh we hosted we hosted some uh presidential hosted we hosted some uh presidential hosted we hosted some uh presidential candidates there and I was like in the candidates there and I was like in the candidates there and I was like in the student government I think I was student student government I think I was student student government I think I was student body and so I had a little lunch with Al body and so I had a little lunch with Al body and so I had a little lunch with Al Gore and uh it was pretty Co did yeah it Gore and uh it was pretty Co did yeah it Gore and uh it was pretty Co did yeah it was nice and he was really down to earth was nice and he was really down to earth was nice and he was really down to earth like he's like a regular person he's not like he's like a regular person he's not like he's like a regular person he's not a yeah he's just just a he's just a a yeah he's just just a he's just a a yeah he's just just a he's just a smart guy he's just a dude yeah that's smart guy he's just a dude yeah that's smart guy he's just a dude yeah that's what most people are just people yeah what most people are just people yeah what most people are just people yeah odd they are you get through all the odd they are you get through all the odd they are you get through all the handlers and the yeah yeah yeah and the handlers and the yeah yeah yeah and the handlers and the yeah yeah yeah and the pr people yeah PR handlers PR handlers

pr people yeah PR handlers PR handlers pr people yeah PR handlers PR handlers yeah but is and they're human yeah he's yeah but is and they're human yeah he's yeah but is and they're human yeah he's the real deal he's the real deal the real deal he's the real deal the real deal he's the real deal in fact I was uh just on a call so one in fact I was uh just on a call so one in fact I was uh just on a call so one of the things um we're doing with tiny of the things um we're doing with tiny of the things um we're doing with tiny ml is we're running some workshops we're ml is we're running some workshops we're ml is we're running some workshops we're actually running a actually running a actually running a sustainability Workshop in Washington DC sustainability Workshop in Washington DC sustainability Workshop in Washington DC in September to talk about Edge in September to talk about Edge in September to talk about Edge Computing and sustainability because as Computing and sustainability because as Computing and sustainability because as you're mentioning it's great you're mentioning it's great you're mentioning it's great intersection there's actually two parts intersection there's actually two parts intersection there's actually two parts to the story I mean there's the one I to the story I mean there's the one I to the story I mean there's the one I would say the what what Rick was posting would say the what what Rick was posting would say the what what Rick was posting this week about you know how all these this week about you know how all these this week about you know how all these Nvidia gpus equal like two 2 million gas Nvidia gpus equal like two 2 million gas Nvidia gpus equal like two 2 million gas powered cars and stuff um but there's powered cars and stuff um but there's powered cars and stuff um but there's the other part too you know the flip the other part too you know the flip the other part too you know the flip side is you know as you proliferate side is you know as you proliferate side is you know as you proliferate devices Edge devices you know there is devices Edge devices you know there is devices Edge devices you know there is an environmental cost for manufacturing an environmental cost for manufacturing an environmental cost for manufacturing those devices and the distribution too those devices and the distribution too those devices and the distribution too so you know that's that's another Factor so you know that's that's another Factor so you know that's that's another Factor we have to keep in mind is like and they we have to keep in mind is like and they we have to keep in mind is like and they and they're also not 100% utilized all and they're also not 100% utilized all and they're also not 100% utilized all the time so you know it's oh yeah the time so you know it's oh yeah the time so you know it's oh yeah definitely not yeah um oh far from 100% definitely not yeah um oh far from 100% definitely not yeah um oh far from 100% utilized if they were then you'll be utilized if they were then you'll be utilized if they were then you'll be walking around with walking around with walking around with devices that you know have like two hour devices that you know have like two hour devices that you know have like two hour well yeah and we all have drawers full well yeah and we all have drawers full well yeah and we all have drawers full of old phones right that are sitting a of old phones right that are sitting a of old phones right that are sitting a lot of lot of lot of E-Waste a lot of E-Waste being generated E-Waste a lot of E-Waste being generated E-Waste a lot of E-Waste being generated on stuff that gets plugged in once or on stuff that gets plugged in once or on stuff that gets plugged in once or it's used for a while and stuff so you it's used for a while and stuff so you it's used for a while and stuff so you know it's it's a complex um calculation know it's it's a complex um calculation know it's it's a complex um calculation um but uh you know it is you know it's

um but uh you know it is you know it's um but uh you know it is you know it's it's the it's there's the obvious things it's the it's there's the obvious things it's the it's there's the obvious things like you look at these you know 1200 like you look at these you know 1200 like you look at these you know 1200 watt black Wells and watt black Wells and watt black Wells and you know putting a building full of them you know putting a building full of them you know putting a building full of them together obviously is going to use a lot together obviously is going to use a lot together obviously is going to use a lot of electricity and power of electricity and power of electricity and power but use power but use power but use power too yeah exactly and I think he's really too yeah exactly and I think he's really too yeah exactly and I think he's really more more more specifically um specifically um specifically um complaining and rightfully so I think complaining and rightfully so I think complaining and rightfully so I think because you know the big question is because you know the big question is because you know the big question is we're doing all this and you know um a we're doing all this and you know um a we're doing all this and you know um a generative AI tools I mean it's generative AI tools I mean it's generative AI tools I mean it's interesting there was a a comment made interesting there was a a comment made interesting there was a a comment made by by a company I forgot the name of it by by a company I forgot the name of it by by a company I forgot the name of it um it's a startup it's been used they um it's a startup it's been used they um it's a startup it's been used they they're using AI to optimize production they're using AI to optimize production they're using AI to optimize production flows and uh you know help in produ like flows and uh you know help in produ like flows and uh you know help in produ like um uh production um uh production um uh production management uh using AI right and this management uh using AI right and this management uh using AI right and this guy went out of his way not to say guy went out of his way not to say guy went out of his way not to say generative generative generative dat H it was interesting dat H it was interesting dat H it was interesting and he must have gotten beat up quite a and he must have gotten beat up quite a and he must have gotten beat up quite a bit over the past year like a lot of bit over the past year like a lot of bit over the past year like a lot of people have like there was a guy here people have like there was a guy here people have like there was a guy here last year you know if you look back on last year you know if you look back on last year you know if you look back on you know I encourage our listeners to you know I encourage our listeners to you know I encourage our listeners to listen back on the episode that we did listen back on the episode that we did listen back on the episode that we did last year about this time where I was last year about this time where I was last year about this time where I was here and one of the things that I here and one of the things that I here and one of the things that I observed was you had some people get up

observed was you had some people get up observed was you had some people get up there and start talk start talking about there and start talk start talking about there and start talk start talking about how generative AI how generative AI how generative AI you know would be able to revolutionize you know would be able to revolutionize you know would be able to revolutionize chip design right and these guys would chip design right and these guys would chip design right and these guys would get lapped off the stage right by get lapped off the stage right by get lapped off the stage right by Engineers who themselves have been you Engineers who themselves have been you Engineers who themselves have been you know tinkering with the tools know tinkering with the tools know tinkering with the tools experimenting and concluding completely experimenting and concluding completely experimenting and concluding completely opposite of what this guy obviously you opposite of what this guy obviously you opposite of what this guy obviously you know when they submitted their topic know when they submitted their topic know when they submitted their topic for um the conference it was done three for um the conference it was done three for um the conference it was done three months months months ago that was three months that engineers ago that was three months that engineers ago that was three months that engineers in the semiconduct industry could in the semiconduct industry could in the semiconduct industry could screwing around with um generative AI screwing around with um generative AI screwing around with um generative AI tools finding out this you really can't tools finding out this you really can't tools finding out this you really can't use this stuff for to the AG that the use this stuff for to the AG that the use this stuff for to the AG that the guy on stage was suggesting right this guy on stage was suggesting right this guy on stage was suggesting right this year you don't year you don't year you don't hear nothing like two sessions on hear nothing like two sessions on hear nothing like two sessions on generative AI right the only place generative AI right the only place generative AI right the only place you're hearing generative AI is in CEO you're hearing generative AI is in CEO you're hearing generative AI is in CEO Su and it's all about demand Su and it's all about demand Su and it's all about demand Supply and Market Supply and Market Supply and Market but that you know those could be but that you know those could be but that you know those could be completely disconnected from what we're completely disconnected from what we're completely disconnected from what we're seeing in terms of actual use within the seeing in terms of actual use within the seeing in terms of actual use within the industry but the IND interesting thing industry but the IND interesting thing industry but the IND interesting thing that he was saying what the one term that he was saying what the one term that he was saying what the one term that he used was transform I did end up that he used was transform I did end up that he used was transform I did end up ask him a bunch of questions transform ask him a bunch of questions transform ask him a bunch of questions transform he was very very cautious yeah about he was very very cautious yeah about he was very very cautious yeah about suggesting this stuff could do anything suggesting this stuff could do anything suggesting this stuff could do anything to the level of exact prediction right

to the level of exact prediction right to the level of exact prediction right like um critical prediction and he he like um critical prediction and he he like um critical prediction and he he ALS said look this is all probabilistic ALS said look this is all probabilistic ALS said look this is all probabilistic it's not deterministic so you know so it's not deterministic so you know so it's not deterministic so you know so all these caveats and obviously there's all these caveats and obviously there's all these caveats and obviously there's a lot of learning going on but when you a lot of learning going on but when you a lot of learning going on but when you look at where we were last year and some look at where we were last year and some look at where we were last year and some of these folks coming in of these folks coming in of these folks coming in impetuously probably having done little impetuously probably having done little impetuously probably having done little homework suggesting Revolution are have homework suggesting Revolution are have homework suggesting Revolution are have completely tempered HW tracks you notice completely tempered HW tracks you notice completely tempered HW tracks you notice it interesting yeah and this is not it interesting yeah and this is not it interesting yeah and this is not uncommon right no I think the pendulum uncommon right no I think the pendulum uncommon right no I think the pendulum is you know pendulum always swings Tech is you know pendulum always swings Tech is you know pendulum always swings Tech world where there's kind of where they world where there's kind of where they world where there's kind of where they say over over estimate the impact in the say over over estimate the impact in the say over over estimate the impact in the in the near term and underestimate it in in the near term and underestimate it in in the near term and underestimate it in the long term uh so yeah you know every the long term uh so yeah you know every the long term uh so yeah you know every a lot of people bring that up but this a lot of people bring that up but this a lot of people bring that up but this is there's one person I F his name is G is there's one person I F his name is G is there's one person I F his name is G kill me if he actually watches the show kill me if he actually watches the show kill me if he actually watches the show says that quite a bit um he's on Jeffrey says that quite a bit um he's on Jeffrey says that quite a bit um he's on Jeffrey funk's threads quite often I said yeah funk's threads quite often I said yeah funk's threads quite often I said yeah you know um the people who subscribe to you know um the people who subscribe to you know um the people who subscribe to the hype right drink deep in it the hype right drink deep in it the hype right drink deep in it they um they overestimate the potential they um they overestimate the potential they um they overestimate the potential in near they also underestimate its in near they also underestimate its in near they also underestimate its potential in the long term because there potential in the long term because there potential in the long term because there there's always this like cycle part of there's always this like cycle part of there's always this like cycle part of this hype cycle is U people getting this hype cycle is U people getting this hype cycle is U people getting their egos bruised for being wrong they their egos bruised for being wrong they their egos bruised for being wrong they either tell a friend to invest or they either tell a friend to invest or they either tell a friend to invest or they suggested hey I have a company that's suggested hey I have a company that's suggested hey I have a company that's going to revolutionize XYZ using XYZ

going to revolutionize XYZ using XYZ going to revolutionize XYZ using XYZ technology sure eating egg on their face technology sure eating egg on their face technology sure eating egg on their face there's denial I mean these people get there's denial I mean these people get there's denial I mean these people get hurt by the height and then they're hurt by the height and then they're hurt by the height and then they're disappointed with the technology because disappointed with the technology because disappointed with the technology because it failed them oh my God it failed them oh my God it failed them oh my God is do this thing how could it betray me is do this thing how could it betray me is do this thing how could it betray me how could how could it not revolutionize how could how could it not revolutionize how could how could it not revolutionize the world and then they're the ones who the world and then they're the ones who the world and then they're the ones who are don't see because they were never are don't see because they were never are don't see because they were never tuned into it in the first place right tuned into it in the first place right tuned into it in the first place right you know what I'm saying and um yeah um you know what I'm saying and um yeah um you know what I'm saying and um yeah um yeah I think it's like the year oh yeah I think it's like the year oh yeah I think it's like the year oh Demetri is coming in the um I think it's Demetri is coming in the um I think it's Demetri is coming in the um I think it's like it has to find its it has to find like it has to find its it has to find like it has to find its it has to find Its Right level it's almost like you Its Right level it's almost like you Its Right level it's almost like you know yeah did you just say hey Demitri know yeah did you just say hey Demitri know yeah did you just say hey Demitri did you say oh did you say oh did you say oh no no no I just I just said oh he's he's no no no I just I just said oh he's he's no no no I just I just said oh he's he's I saw the little butt oh my God I of I saw the little butt oh my God I of I saw the little butt oh my God I of course he looks like he's in a nice scho course he looks like he's in a nice scho course he looks like he's in a nice scho love youe he's in a cool environment love youe he's in a cool environment love youe he's in a cool environment he's got some nice is that Southport sou he's got some nice is that Southport sou he's got some nice is that Southport sou yeah I'm student of the South we yeah I'm student of the South we yeah I'm student of the South we talk about oh my talk about oh my talk about oh my God I love God I love God I love it next thing you have to put a it next thing you have to put a it next thing you have to put a idiocracy background there you go in the idiocracy background there you go in the idiocracy background there you go in the background yeah where they where they background yeah where they where they background yeah where they where they watering where they watering plants with watering where they watering plants with watering where they watering plants with that electrolyte drink what was called that electrolyte drink what was called that electrolyte drink what was called you are at another conference using yeah you are at another conference using yeah you are at another conference using yeah I'm in San I'm in San I'm in San Francisco semicon West yeah MOS

Francisco semicon West yeah MOS Francisco semicon West yeah MOS and um yeah you know the other thing um and um yeah you know the other thing um and um yeah you know the other thing um Pete and ditri this year it seems a Pete and ditri this year it seems a Pete and ditri this year it seems a little bit mellow I heard that there little bit mellow I heard that there little bit mellow I heard that there were more people this year but there were more people this year but there were more people this year but there isn't as much of a you know what I'm isn't as much of a you know what I'm isn't as much of a you know what I'm saying energy Buzz to the yeah less of saying energy Buzz to the yeah less of saying energy Buzz to the yeah less of an energy buzz and you know even with an energy buzz and you know even with an energy buzz and you know even with this sustainability stuff it's like this sustainability stuff it's like this sustainability stuff it's like people just don't seem to care as much people just don't seem to care as much people just don't seem to care as much last year it was big on sustainability last year it was big on sustainability last year it was big on sustainability now it's like do we give up on it is the now it's like do we give up on it is the now it's like do we give up on it is the COO Summit yeah COO Summit yeah COO Summit yeah yeah they're probably all hiding because yeah they're probably all hiding because yeah they're probably all hiding because of the layoffs they had to do in the bus of the layoffs they had to do in the bus of the layoffs they had to do in the bus thinking about how to raise money I thinking about how to raise money I thinking about how to raise money I found a bunch of folks like when we had found a bunch of folks like when we had found a bunch of folks like when we had our event in Milan recently there's a our event in Milan recently there's a our event in Milan recently there's a lot of these still a lot of lot of these still a lot of lot of these still a lot of new um I would say like neuromorphic new um I would say like neuromorphic new um I would say like neuromorphic kind of AI chip yeah companies coming up kind of AI chip yeah companies coming up kind of AI chip yeah companies coming up you know that are getting funding series you know that are getting funding series you know that are getting funding series a like doing like spiking neural a like doing like spiking neural a like doing like spiking neural networks and you know um talk to a networks and you know um talk to a networks and you know um talk to a company yesterday company yesterday company yesterday yeah it's like all this stuff for like yeah it's like all this stuff for like yeah it's like all this stuff for like brain computer interface and so I think brain computer interface and so I think brain computer interface and so I think we're going to see a lot of interesting we're going to see a lot of interesting we're going to see a lot of interesting silicon there I you're not going to drop silicon there I you're not going to drop silicon there I you're not going to drop like a Intel Core Ultra into your into like a Intel Core Ultra into your into like a Intel Core Ultra into your into your Noggin uh but uh you could drop in your Noggin uh but uh you could drop in your Noggin uh but uh you could drop in you know an implant to do deep brain you know an implant to do deep brain you know an implant to do deep brain stimulation that that has a little bit stimulation that that has a little bit stimulation that that has a little bit of an AI capability on it to to monitor of an AI capability on it to to monitor of an AI capability on it to to monitor your brain signals so I think there's a your brain signals so I think there's a your brain signals so I think there's a ton of interesting stuff happening there ton of interesting stuff happening there ton of interesting stuff happening there with new silicon companies that are with new silicon companies that are with new silicon companies that are designing specifically for purpose like designing specifically for purpose like designing specifically for purpose like not General it's specifically silicon

not General it's specifically silicon not General it's specifically silicon for the healthcare industry or whatever for the healthcare industry or whatever for the healthcare industry or whatever and I think some of the new and I think some of the new and I think some of the new techniques and Fab and other things that techniques and Fab and other things that techniques and Fab and other things that you were talking about are enabling much you were talking about are enabling much you were talking about are enabling much more specialized silicon development for more specialized silicon development for more specialized silicon development for these verticals um so I think that's these verticals um so I think that's these verticals um so I think that's kind of exciting and and it's an kind of exciting and and it's an kind of exciting and and it's an opportunity for a lot of startups to you opportunity for a lot of startups to you opportunity for a lot of startups to you know find their Lane and really yeah know find their Lane and really yeah know find their Lane and really yeah nail that particular vertical not try to nail that particular vertical not try to nail that particular vertical not try to you know the traditional semiconductor you know the traditional semiconductor you know the traditional semiconductor approach right as you build a platform approach right as you build a platform approach right as you build a platform and it's general purpose and you sell and it's general purpose and you sell and it's general purpose and you sell the out of it and it's distribut the out of it and it's distribut the out of it and it's distribut these days it feels like it's all very these days it feels like it's all very these days it feels like it's all very designed for for purpose from the designed for for purpose from the designed for for purpose from the beginning but I don't know we'll see beginning but I don't know we'll see beginning but I don't know we'll see yeah I I think you're right I mean there yeah I I think you're right I mean there yeah I I think you're right I mean there is a lot more talk about that now trying is a lot more talk about that now trying is a lot more talk about that now trying um so as some of these large language um so as some of these large language um so as some of these large language models get bigger and become a larger models get bigger and become a larger models get bigger and become a larger problem to solve but are very specific problem to solve but are very specific problem to solve but are very specific to a particular algorithm or model set to a particular algorithm or model set to a particular algorithm or model set um there are companies out here that are um there are companies out here that are um there are companies out here that are building as6 and all and also very building as6 and all and also very building as6 and all and also very bespoked software to optimally run these bespoked software to optimally run these bespoked software to optimally run these things at a much higher well what they things at a much higher well what they things at a much higher well what they claim is a much higher per per wat um uh claim is a much higher per per wat um uh claim is a much higher per per wat um uh you you you know Advantage than than um general know Advantage than than um general know Advantage than than um general purpose period yeah General general purpose period yeah General general purpose period yeah General general purpose purpose purpose GPU yes General verp well they are I GPU yes General verp well they are I GPU yes General verp well they are I mean it's and you know and it's the as mean it's and you know and it's the as mean it's and you know and it's the as people find use cases they're building people find use cases they're building people find use cases they're building sulur and silicon for that use case yeah

sulur and silicon for that use case yeah sulur and silicon for that use case yeah yeah um yeah and whether it's generative yeah um yeah and whether it's generative yeah um yeah and whether it's generative AI Transformer based models or more AI Transformer based models or more AI Transformer based models or more machine learning it's very bespoke and machine learning it's very bespoke and machine learning it's very bespoke and focused these days for particular yeah focused these days for particular yeah focused these days for particular yeah and so if the problems are big enough and so if the problems are big enough and so if the problems are big enough then there's an Asic that can solve a then there's an Asic that can solve a then there's an Asic that can solve a problem you know um and I think one of problem you know um and I think one of problem you know um and I think one of the big the big the big misconceptions you know because I've misconceptions you know because I've misconceptions you know because I've been talking to a lot of folks here been talking to a lot of folks here been talking to a lot of folks here there is there is there is um long sustained argument now and you um long sustained argument now and you um long sustained argument now and you hear a lot of analysts and others talk hear a lot of analysts and others talk hear a lot of analysts and others talk about this is how software has been about this is how software has been about this is how software has been nvidia's Advantage I don't think so I nvidia's Advantage I don't think so I nvidia's Advantage I don't think so I think what's actually been nvidia's think what's actually been nvidia's think what's actually been nvidia's Advantage has been interconnect and Advantage has been interconnect and Advantage has been interconnect and networking that's what's made the big networking that's what's made the big networking that's what's made the big difference as soon as you solve that the difference as soon as you solve that the difference as soon as you solve that the software part Cuda and all this stuff is software part Cuda and all this stuff is software part Cuda and all this stuff is going to matter less a lot less is that going to matter less a lot less is that going to matter less a lot less is that the Millan and the Millan and the Millan and yeah the melanox stuff the yeah so and yeah the melanox stuff the yeah so and yeah the melanox stuff the yeah so and their own Envy link so melanox is going their own Envy link so melanox is going their own Envy link so melanox is going to be the infin ban you know like the to be the infin ban you know like the to be the infin ban you know like the the high higher level um you know um um the high higher level um you know um um the high higher level um you know um um server to server server to server server to server RoR uh interconnect and then you have RoR uh interconnect and then you have RoR uh interconnect and then you have like Envy link which is like between like Envy link which is like between like Envy link which is like between gpus on a server right and so you have gpus on a server right and so you have gpus on a server right and so you have that high bandwidth high capacity that high bandwidth high capacity that high bandwidth high capacity interconnect that interconnect that interconnect that um which is has incredibly low um which is has incredibly low um which is has incredibly low latency and you know so you have UA link

latency and you know so you have UA link latency and you know so you have UA link and the ultra EET groups trying to to and the ultra EET groups trying to to and the ultra EET groups trying to to bring something to the table that could bring something to the table that could bring something to the table that could ququ be ququ be ququ be standardized um that can compete right standardized um that can compete right standardized um that can compete right with with with um um with nvidia's interconnect um um with nvidia's interconnect um um with nvidia's interconnect offerings and you know that's obviously offerings and you know that's obviously offerings and you know that's obviously the the advance AG that theid has over the the advance AG that theid has over the the advance AG that theid has over everything else so right right um yeah everything else so right right um yeah everything else so right right um yeah and then you know there's all the and then you know there's all the and then you know there's all the platform stuff that uh Nvidia always platform stuff that uh Nvidia always platform stuff that uh Nvidia always talks about like the Omniverse and yada talks about like the Omniverse and yada talks about like the Omniverse and yada yada but when you you look at the other yada but when you you look at the other yada but when you you look at the other parts of their um portfolio they're not parts of their um portfolio they're not parts of their um portfolio they're not as as as strong um those those adoption of strong um those those adoption of strong um those those adoption of Omniverse and and um the other aspects Omniverse and and um the other aspects Omniverse and and um the other aspects of their has not thing as high as people of their has not thing as high as people of their has not thing as high as people think you know I go to different think you know I go to different think you know I go to different companies I ask folks uh you know are companies I ask folks uh you know are companies I ask folks uh you know are you guys using Omniverse for digital you guys using Omniverse for digital you guys using Omniverse for digital twin and blah blah blah and simulation twin and blah blah blah and simulation twin and blah blah blah and simulation because Nvidia talks about all the time because Nvidia talks about all the time because Nvidia talks about all the time they actually don't and then a lot of they actually don't and then a lot of they actually don't and then a lot of the isbs and the tech vendors in the isbs and the tech vendors in the isbs and the tech vendors in industrial have their own tool sets that industrial have their own tool sets that industrial have their own tool sets that have been around forever tool sets are have been around forever tool sets are have been around forever tool sets are pretty whether it's seens or you know pretty whether it's seens or you know pretty whether it's seens or you know doel or whoever and and they're just as doel or whoever and and they're just as doel or whoever and and they're just as good good good and they have they're as standardized and they have they're as standardized and they have they're as standardized within um to a certain degree um but

within um to a certain degree um but within um to a certain degree um but yeah man it's um the Omniverse thing is yeah man it's um the Omniverse thing is yeah man it's um the Omniverse thing is interesting because I think it's it interesting because I think it's it interesting because I think it's it sounds great I think it's just the you sounds great I think it's just the you sounds great I think it's just the you know right now like I can get an Orin know right now like I can get an Orin know right now like I can get an Orin you know box and Cuda and I can be up you know box and Cuda and I can be up you know box and Cuda and I can be up and running in an hour if I want to use and running in an hour if I want to use and running in an hour if I want to use Omniverse for my business that's like a Omniverse for my business that's like a Omniverse for my business that's like a very big investment to do a lot of very big investment to do a lot of very big investment to do a lot of simulation and modeling and tuning and simulation and modeling and tuning and simulation and modeling and tuning and so you need to bring in a system so you need to bring in a system so you need to bring in a system integrator design you know it's like a integrator design you know it's like a integrator design you know it's like a whole thing and so it's pretty big yeah whole thing and so it's pretty big yeah whole thing and so it's pretty big yeah a big thing and for big companies and a big thing and for big companies and a big thing and for big companies and big factories and stuff I totally get it big factories and stuff I totally get it big factories and stuff I totally get it it's just it's just not a scale thing it's just it's just not a scale thing it's just it's just not a scale thing just like you know talk about humanoid just like you know talk about humanoid just like you know talk about humanoid robots I mean they're really cool they robots I mean they're really cool they robots I mean they're really cool they make good demos but in terms of like make good demos but in terms of like make good demos but in terms of like unit volume of humanoid robots you know unit volume of humanoid robots you know unit volume of humanoid robots you know you're talking about maybe a million you're talking about maybe a million you're talking about maybe a million units 10 years from now or something units 10 years from now or something units 10 years from now or something like that it's not it's just not a big like that it's not it's just not a big like that it's not it's just not a big scale play and so it's expensive and scale play and so it's expensive and scale play and so it's expensive and there's a lot of lot of upfront there's a lot of lot of upfront there's a lot of lot of upfront investment to make this stuff work so I investment to make this stuff work so I investment to make this stuff work so I think some of these newer Technologies I think some of these newer Technologies I think some of these newer Technologies I mean going back to our other thing you mean going back to our other thing you mean going back to our other thing you know we kind of get really excited and know we kind of get really excited and know we kind of get really excited and breathless about them but takes a long breathless about them but takes a long breathless about them but takes a long time for them to get any traction and to time for them to get any traction and to time for them to get any traction and to get any scale so cool stuff just you get any scale so cool stuff just you get any scale so cool stuff just you know it's like pretty you know it's kind know it's like pretty you know it's kind know it's like pretty you know it's kind of Double Black Diamond degree of Double Black Diamond degree of Double Black Diamond degree difficulty to implement and deploy yeah difficulty to implement and deploy yeah difficulty to implement and deploy yeah double black I like that you gota you double black I like that you gota you double black I like that you gota you got to hit all the the what do you call got to hit all the the what do you call got to hit all the the what do you call it the it the it the what do they call it the the green what do they call it the the green what do they call it the the green circle that's right the beginner run circle that's right the beginner run circle that's right the beginner run stuff I'm inspired by and maybe

stuff I'm inspired by and maybe stuff I'm inspired by and maybe maybe start with the Bunny Hill and then maybe start with the Bunny Hill and then maybe start with the Bunny Hill and then uh work your way up oh my God Dimitri uh work your way up oh my God Dimitri uh work your way up oh my God Dimitri it's odd that you're quiet you haven't it's odd that you're quiet you haven't it's odd that you're quiet you haven't said much no this thing I'm learning said much no this thing I'm learning said much no this thing I'm learning sometimes oh they talk about digital sometimes oh they talk about digital sometimes oh they talk about digital twin here dude yeah actually I never twin here dude yeah actually I never twin here dude yeah actually I never looked really exactly at what what Nidia looked really exactly at what what Nidia looked really exactly at what what Nidia has been doing with their Multiverse that's like you know would Universe that's like you know would Universe wouldn't there be some concern about wouldn't there be some concern about wouldn't there be some concern about data privacy around that because it's a data privacy around that because it's a data privacy around that because it's a cloud offering right so if as a cloud offering right so if as a cloud offering right so if as a manufacturer I decided to put all my manufacturer I decided to put all my manufacturer I decided to put all my stuff in simulation there the data is is stuff in simulation there the data is is stuff in simulation there the data is is in their Camp right yeah I mean they in their Camp right yeah I mean they in their Camp right yeah I mean they have the usual kind of confidential have the usual kind of confidential have the usual kind of confidential Computing type of things stuff going on Computing type of things stuff going on Computing type of things stuff going on a wrapid all that stuff but because that a wrapid all that stuff but because that a wrapid all that stuff but because that was a that was a big deal when when I was a that was a big deal when when I was a that was a big deal when when I was at G the the whole the customer say was at G the the whole the customer say was at G the the whole the customer say oh yeah yeah it's a cloud yeah but it's oh yeah yeah it's a cloud yeah but it's oh yeah yeah it's a cloud yeah but it's your Cloud you know so can I have a your Cloud you know so can I have a your Cloud you know so can I have a private Cloud version private Cloud version private Cloud version so so it's it's true but I think your so so it's it's true but I think your so so it's it's true but I think your your point also that most of the large your point also that most of the large your point also that most of the large vendors already have their their vendors already have their their vendors already have their their solution is solution is solution is there in general in in in the in there in general in in in the in there in general in in in the in industrial iot I I mean I don't think industrial iot I I mean I don't think industrial iot I I mean I don't think I've seen any evidence of change in in I've seen any evidence of change in in I've seen any evidence of change in in my position there if you're looking at my position there if you're looking at my position there if you're looking at Big assets like you know gas gas Big assets like you know gas gas Big assets like you know gas gas turbines or jet engin or big turbines or jet engin or big turbines or jet engin or big manufacturer they're going to come up manufacturer they're going to come up manufacturer they're going to come up with a software solution that is with a software solution that is with a software solution that is proprietary and serve the purpose the

proprietary and serve the purpose the proprietary and serve the purpose the the the problem is in the the pumps the the the problem is in the the pumps the the the problem is in the the pumps the compressors so all the manufacturers compressors so all the manufacturers compressors so all the manufacturers those smaller assets who cannot have the those smaller assets who cannot have the those smaller assets who cannot have the the investment into cannot the the investment into cannot the the investment into cannot the investment own platform and cannot find investment own platform and cannot find investment own platform and cannot find platform because a go G's a stuck in G's platform because a go G's a stuck in G's platform because a go G's a stuck in G's a Go Cement stuck in cement so it's a Go Cement stuck in cement so it's a Go Cement stuck in cement so it's problem of the middle ground problem of the middle ground problem of the middle ground that nobody wants to solve because that nobody wants to solve because that nobody wants to solve because everybody's AF well this is also the everybody's AF well this is also the everybody's AF well this is also the Green Field Brownfield issue it's Green Field Brownfield issue it's Green Field Brownfield issue it's chronic issue right I mean fact is 90% chronic issue right I mean fact is 90% chronic issue right I mean fact is 90% of the stuff is Brownfield there's stuff of the stuff is Brownfield there's stuff of the stuff is Brownfield there's stuff out there it's working you know what out there it's working you know what out there it's working you know what they say if it's not broke don't fix it they say if it's not broke don't fix it they say if it's not broke don't fix it so the last thing you want to do is you so the last thing you want to do is you so the last thing you want to do is you know rip rip and replace with a green know rip rip and replace with a green know rip rip and replace with a green field solution and and so there so much field solution and and so there so much field solution and and so there so much effort put into uh even recently we're effort put into uh even recently we're effort put into uh even recently we're talking to folks about I think it was talking to folks about I think it was talking to folks about I think it was like Electrolux talking about how they like Electrolux talking about how they like Electrolux talking about how they do more ml now on their current draw of do more ml now on their current draw of do more ml now on their current draw of their washers and dryers and things like their washers and dryers and things like their washers and dryers and things like that so just from the current draw that so just from the current draw that so just from the current draw itself you can look at anomaly detection itself you can look at anomaly detection itself you can look at anomaly detection and use AI to determine whether the and use AI to determine whether the and use AI to determine whether the dryer is overloaded or whatever and yeah dryer is overloaded or whatever and yeah dryer is overloaded or whatever and yeah I think there's a good I actually saw I think there's a good I actually saw I think there's a good I actually saw something at tin ml the T conference in something at tin ml the T conference in something at tin ml the T conference in in in Burling when you as well pit I in in Burling when you as well pit I in in Burling when you as well pit I think there's a lot of on the brown Fe think there's a lot of on the brown Fe think there's a lot of on the brown Fe aspect there's opportunity forrit so if aspect there's opportunity forrit so if aspect there's opportunity forrit so if you can put a smalll module in line with you can put a smalll module in line with you can put a smalll module in line with the current conception motor and kind of the current conception motor and kind of the current conception motor and kind of stuff in Gaza yeah you can do it all the stuff in Gaza yeah you can do it all the stuff in Gaza yeah you can do it all the motor itself is the sensor basically the motor itself is the sensor basically the motor itself is the sensor basically the motor is the sensor and uh there's a lot motor is the sensor and uh there's a lot motor is the sensor and uh there's a lot of information you can get off the motor

of information you can get off the motor of information you can get off the motor from a retrofit perspective and and do a from a retrofit perspective and and do a from a retrofit perspective and and do a lot with that information so yeah I mean lot with that information so yeah I mean lot with that information so yeah I mean you know and again at the end of the day you know and again at the end of the day you know and again at the end of the day when you're on the buy side of it you when you're on the buy side of it you when you're on the buy side of it you know a lot of folks are on the supply know a lot of folks are on the supply know a lot of folks are on the supply side making Tech but when you're on the side making Tech but when you're on the side making Tech but when you're on the buy side you know you start with a buy side you know you start with a buy side you know you start with a problem and you work your way back to problem and you work your way back to problem and you work your way back to how quickly and cheaply can I fix it um how quickly and cheaply can I fix it um how quickly and cheaply can I fix it um times that means just you know measuring times that means just you know measuring times that means just you know measuring current draw on your your current draw on your your current draw on your your motor yeah it's the same I mean it's the motor yeah it's the same I mean it's the motor yeah it's the same I mean it's the same level of the value of the assets same level of the value of the assets same level of the value of the assets because you got the same problem you because you got the same problem you because you got the same problem you know if the motor is inexpensive you're know if the motor is inexpensive you're know if the motor is inexpensive you're probably going to buy the new y with the probably going to buy the new y with the probably going to buy the new y with the AI and the sensors in the ml but if it's AI and the sensors in the ml but if it's AI and the sensors in the ml but if it's a big as expensive one you might want to a big as expensive one you might want to a big as expensive one you might want to put sensor on it then so this balance of put sensor on it then so this balance of put sensor on it then so this balance of do I throw away replace or do I try to do I throw away replace or do I try to do I throw away replace or do I try to retrofit and add some stuff it well I retrofit and add some stuff it well I retrofit and add some stuff it well I guess going back to the sustainability guess going back to the sustainability guess going back to the sustainability thing I think the idea is don't throw it thing I think the idea is don't throw it thing I think the idea is don't throw it away exactly and the result also of that away exactly and the result also of that away exactly and the result also of that is that just Chuck it in the garbage go is that just Chuck it in the garbage go is that just Chuck it in the garbage go the result also of this is H at the end the result also of this is H at the end the result also of this is H at the end of the day most of the industrial use of the day most of the industrial use of the day most of the industrial use cases are system integration projects oh cases are system integration projects oh cases are system integration projects oh yeah why to some extent you the software yeah why to some extent you the software yeah why to some extent you the software part or the the digital twin part part or the the digital twin part part or the the digital twin part nobody's really motivated by that nobody's really motivated by that nobody's really motivated by that because they want to keep theer in because they want to keep theer in because they want to keep theer in integrating the system setting it isn't integrating the system setting it isn't integrating the system setting it isn't isn't alltech these days even cons Tech isn't alltech these days even cons Tech isn't alltech these days even cons Tech a system integration project I mean you a system integration project I mean you a system integration project I mean you know you add a doorbell to your house know you add a doorbell to your house know you add a doorbell to your house and then you got an Alexa skill and then and then you got an Alexa skill and then and then you got an Alexa skill and then the thing and like you know every you the thing and like you know every you the thing and like you know every you add a new sprinkler system and all of a add a new sprinkler system and all of a add a new sprinkler system and all of a sudden it's like you're you're cracking

sudden it's like you're you're cracking sudden it's like you're you're cracking open your whole uh home infrastructure open your whole uh home infrastructure open your whole uh home infrastructure to sort of get this thing to work with to sort of get this thing to work with to sort of get this thing to work with all your other googas so everything's a all your other googas so everything's a all your other googas so everything's a system integration problem yeah home system integration problem yeah home system integration problem yeah home automation is a nightmare unless you automation is a nightmare unless you automation is a nightmare unless you select one vendor say okay I'm going to select one vendor say okay I'm going to select one vendor say okay I'm going to go do everything with apple and then you go do everything with apple and then you go do everything with apple and then you might get something otherwise just it's might get something otherwise just it's might get something otherwise just it's a nightmare the home kit I actually was a nightmare the home kit I actually was a nightmare the home kit I actually was trying to retrofit I have a Chamberlain trying to retrofit I have a Chamberlain trying to retrofit I have a Chamberlain uh garage door opener with this wi-fi uh garage door opener with this wi-fi uh garage door opener with this wi-fi thing but you can't use hom kit with it thing but you can't use hom kit with it thing but you can't use hom kit with it like it won't you can't verbally say like it won't you can't verbally say like it won't you can't verbally say open and close because they think this open and close because they think this open and close because they think this is a security risk which makes no sense is a security risk which makes no sense is a security risk which makes no sense especially my garage which is detached especially my garage which is detached especially my garage which is detached but whatever um but they won't let me do but whatever um but they won't let me do but whatever um but they won't let me do it so there's a third party box you can it so there's a third party box you can it so there's a third party box you can get that's like a homekit add-on that get that's like a homekit add-on that get that's like a homekit add-on that kind of hacks like access a remote for kind of hacks like access a remote for kind of hacks like access a remote for your remote to hack a homekit uh your remote to hack a homekit uh your remote to hack a homekit uh commands into your garage door opener so commands into your garage door opener so commands into your garage door opener so you could say Hey Siri open my garage you could say Hey Siri open my garage you could say Hey Siri open my garage but that's like a third party box and but that's like a third party box and but that's like a third party box and it's nuts it's insane to try to hook all it's nuts it's insane to try to hook all it's nuts it's insane to try to hook all this stuff up but that's that's a the this stuff up but that's that's a the this stuff up but that's that's a the argument that you know opening the argument that you know opening the argument that you know opening the things make it insecure is just the things make it insecure is just the things make it insecure is just the argument for them to sell you their argument for them to sell you their argument for them to sell you their platform and the fact that they're not platform and the fact that they're not platform and the fact that they're not capable of Designing it properly no yeah capable of Designing it properly no yeah capable of Designing it properly no yeah I mean no they don't even have an option I mean no they don't even have an option I mean no they don't even have an option they just don't do it and it's probably they just don't do it and it's probably they just don't do it and it's probably because they just didn't figure it out because they just didn't figure it out because they just didn't figure it out and whatever the intern for the summer and whatever the intern for the summer and whatever the intern for the summer left went back to school and they don't left went back to school and they don't left went back to school and they don't have the code so that's the that's the have the code so that's the that's the have the code so that's the that's the how it works right the the internet how it works right the the internet how it works right the the internet chamber left or or he wrote the code and chamber left or or he wrote the code and chamber left or or he wrote the code and the viable names are well first names the viable names are well first names the viable names are well first names yeah I know someone that did that in the

yeah I know someone that did that in the yeah I know someone that did that in the past he thought he past he thought he past he thought he was so if anyone from Chamberlain's out was so if anyone from Chamberlain's out was so if anyone from Chamberlain's out there hire another intern that's for there hire another intern that's for there hire another intern that's for sure you get that fixed too funny yeah so I you get that fixed too funny yeah so I haven't watched the news in the past haven't watched the news in the past haven't watched the news in the past week so anything new in the iot space week so anything new in the iot space week so anything new in the iot space that is that is that is exciting iot space new um well we're exciting iot space new um well we're exciting iot space new um well we're were just talking about Rick's uh rant were just talking about Rick's uh rant were just talking about Rick's uh rant before you got on here about yeah I before you got on here about yeah I before you got on here about yeah I mean I don't nice apocalyptic Vision mean I don't nice apocalyptic Vision mean I don't nice apocalyptic Vision that he hase with him if if we continue that he hase with him if if we continue that he hase with him if if we continue with this overflow of AI stupidity we're with this overflow of AI stupidity we're with this overflow of AI stupidity we're going to burn so much energy trying going to burn so much energy trying going to burn so much energy trying Leonard was saying we're we're you know Leonard was saying we're we're you know Leonard was saying we're we're you know the pendulum is starting to just like the pendulum is starting to just like the pendulum is starting to just like any other piece of tech this you know any other piece of tech this you know any other piece of tech this you know the pendulum is swinging back I was the pendulum is swinging back I was the pendulum is swinging back I was reading too speaking a pendulum swinging reading too speaking a pendulum swinging reading too speaking a pendulum swinging back I was reading today about uh some back I was reading today about uh some back I was reading today about uh some analysis on the aipc and how qualcomm's analysis on the aipc and how qualcomm's analysis on the aipc and how qualcomm's kind of shifted their value prop to be a kind of shifted their value prop to be a kind of shifted their value prop to be a little more about battery life first and little more about battery life first and little more about battery life first and then sort of AI is like a cool little then sort of AI is like a cool little then sort of AI is like a cool little booster but you know the value prop for booster but you know the value prop for booster but you know the value prop for buying and this was back when I was buying and this was back when I was buying and this was back when I was working on Windows and Snapdragon 2 it's working on Windows and Snapdragon 2 it's working on Windows and Snapdragon 2 it's like it's just killer battery it's just like it's just killer battery it's just like it's just killer battery it's just killer battery life and like that's why killer battery life and like that's why killer battery life and like that's why people when people first started putting people when people first started putting people when people first started putting batteries in PCS back in the 80s and 90s batteries in PCS back in the 80s and 90s batteries in PCS back in the 80s and 90s the value prop you could take it with the value prop you could take it with the value prop you could take it with you and use it yeah and that that you and use it yeah and that that you and use it yeah and that that battery life has always been the main battery life has always been the main battery life has always been the main value yeah you could take it on the value yeah you could take it on the value yeah you could take it on the plane and get about two hours of battery plane and get about two hours of battery plane and get about two hours of battery life and that was pretty good yeah well life and that was pretty good yeah well life and that was pretty good yeah well but I mean yeah to Rick's argument you

but I mean yeah to Rick's argument you but I mean yeah to Rick's argument you know we are in a situation where the know we are in a situation where the know we are in a situation where the number of owner of magnitudes is much number of owner of magnitudes is much number of owner of magnitudes is much much much different I mean we seeking much much different I mean we seeking much much different I mean we seeking general intelligence general general intelligence general general intelligence general intelligence here 300 watts 300 watts intelligence here 300 watts 300 watts intelligence here 300 watts 300 watts think about how much we are we're trying think about how much we are we're trying think about how much we are we're trying to to to so yeah yeah well you know there's also so yeah yeah well you know there's also so yeah yeah well you know there's also far away so and again the models in far away so and again the models in far away so and again the models in current AI have some applicability I do current AI have some applicability I do current AI have some applicability I do believe but we're going to reach a point believe but we're going to reach a point believe but we're going to reach a point where given the the type of Technology where given the the type of Technology where given the the type of Technology we use the energy consumption is such we use the energy consumption is such we use the energy consumption is such that it's or it's going to be just for a that it's or it's going to be just for a that it's or it's going to be just for a few elit and maybe the the dream of you few elit and maybe the the dream of you few elit and maybe the the dream of you know there only a market for five or six know there only a market for five or six know there only a market for five or six computers that IBM has that there will computers that IBM has that there will computers that IBM has that there will only be a market for five of six gen I only be a market for five of six gen I only be a market for five of six gen I machines because the amount of energy machines because the amount of energy machines because the amount of energy they consume is so high that only few they consume is so high that only few they consume is so high that only few people could afford it yeah there may be people could afford it yeah there may be people could afford it yeah there may be some a few few companies that have some a few few companies that have some a few few companies that have Uber um model not everyone's G have one Uber um model not everyone's G have one Uber um model not everyone's G have one um and everything else is going to be um and everything else is going to be um and everything else is going to be just a reduce you know they're they're just a reduce you know they're they're just a reduce you know they're they're talking more here about small language talking more here about small language talking more here about small language model so hey I ml um they're talking model so hey I ml um they're talking model so hey I ml um they're talking more about like reduced precision more about like reduced precision more about like reduced precision um all these things that have nothing to um all these things that have nothing to um all these things that have nothing to do with silic um it's just the models do with silic um it's just the models do with silic um it's just the models themselves more uh compressed and themselves more uh compressed and themselves more uh compressed and Deployable on a smaller footprint of Deployable on a smaller footprint of Deployable on a smaller footprint of Hardware um that there's starting to

Hardware um that there's starting to Hardware um that there's starting to talk about that yeah here they didn't do talk about that yeah here they didn't do talk about that yeah here they didn't do that last year because they were all that last year because they were all that last year because they were all focused on this massive competing stuff focused on this massive competing stuff focused on this massive competing stuff and there you know some people are and there you know some people are and there you know some people are starting to realize oh you know this starting to realize oh you know this starting to realize oh you know this stuff isn't really stuff isn't really stuff isn't really AI this is just superc AI this is just superc AI this is just superc Computing right and then you know Computing right and then you know Computing right and then you know there's a couple of slides that W up there's a couple of slides that W up there's a couple of slides that W up like yeah you know um this is just about like yeah you know um this is just about like yeah you know um this is just about superc Computing and Nvidia doesn't have superc Computing and Nvidia doesn't have superc Computing and Nvidia doesn't have the largest super computers what they're the largest super computers what they're the largest super computers what they're doing is they're kind of semi doing is they're kind of semi doing is they're kind of semi democratizing it a bit and making it a democratizing it a bit and making it a democratizing it a bit and making it a general purpose right um but yeah it's general purpose right um but yeah it's general purpose right um but yeah it's that's all this is It's you know when that's all this is It's you know when that's all this is It's you know when you look at the nuts and bolts what's you look at the nuts and bolts what's you look at the nuts and bolts what's happening really it's super confuting happening really it's super confuting happening really it's super confuting the AI stuff yeah the super Computing the AI stuff yeah the super Computing the AI stuff yeah the super Computing enabled Transformers that you probably enabled Transformers that you probably enabled Transformers that you probably could have run this on other other super could have run this on other other super could have run this on other other super computers it's just that they weren't as computers it's just that they weren't as computers it's just that they weren't as accessible they're also not as accessible they're also not as accessible they're also not as extensible or flexible to extensible or flexible to extensible or flexible to support and not optimize for support and not optimize for support and not optimize for transformance even though I think the transformance even though I think the transformance even though I think the early Nvidia systems also weren't early Nvidia systems also weren't early Nvidia systems also weren't optimized for Trans optimized for Trans optimized for Trans they will tell you well we had we had it they will tell you well we had we had it they will tell you well we had we had it kind of tun for kind of tun for kind of tun for Transformers okay but they made some Transformers okay but they made some Transformers okay but they made some huge improvements you know they have a huge improvements you know they have a huge improvements you know they have a Transformer Transformer Transformer engine um and a number of other tools engine um and a number of other tools engine um and a number of other tools they' done to the new newer systems like

they' done to the new newer systems like they' done to the new newer systems like the B and the B and the B and the the hopper systems that the the hopper systems that the the hopper systems that uh you know they're better suited than uh you know they're better suited than uh you know they're better suited than any other computer but that any other computer but that any other computer but that doesn't do you know what I'm say doesn't do you know what I'm say doesn't do you know what I'm say no they no they no they necessarily own the crown for super necessarily own the crown for super necessarily own the crown for super Computing they they're definitely Computing they they're definitely Computing they they're definitely um yeah I guess you could say disrupting um yeah I guess you could say disrupting um yeah I guess you could say disrupting to a certain to a certain to a certain degree but they're also taking like degree but they're also taking like degree but they're also taking like you're saying before taking a very you're saying before taking a very you're saying before taking a very holistic approach to it with the holistic approach to it with the holistic approach to it with the interconnects and the interconnects and the interconnects and the Silicon um and kind of putting the whole Silicon um and kind of putting the whole Silicon um and kind of putting the whole package together the other thing that package together the other thing that package together the other thing that most people maybe are not aware of is most people maybe are not aware of is most people maybe are not aware of is you know when you're run running you know when you're run running you know when you're run running Transformer based models it's as much Transformer based models it's as much Transformer based models it's as much about the memory as it is about the about the memory as it is about the about the memory as it is about the horsepower of the the chip so the memory horsepower of the the chip so the memory horsepower of the the chip so the memory usage is huge and so that's you know the usage is huge and so that's you know the usage is huge and so that's you know the first company that can sort of figure first company that can sort of figure first company that can sort of figure out how to radically reduce the amount out how to radically reduce the amount out how to radically reduce the amount of memory you need to run because we're of memory you need to run because we're of memory you need to run because we're running like small language models on running like small language models on running like small language models on all kinds of things here at tiny ML and all kinds of things here at tiny ML and all kinds of things here at tiny ML and the the biggest issue is not the chip the the biggest issue is not the chip the the biggest issue is not the chip horsepower it's the memory and the horsepower it's the memory and the horsepower it's the memory and the memory cost yeah so you know that's memory cost yeah so you know that's memory cost yeah so you know that's architecturally just a lot you have to architecturally just a lot you have to architecturally just a lot you have to have a lot of stuff in play at the same have a lot of stuff in play at the same have a lot of stuff in play at the same time in memory to run these models and time in memory to run these models and time in memory to run these models and uh solving that will be huge because uh solving that will be huge because uh solving that will be huge because then you really start to see these more then you really start to see these more then you really start to see these more kind of Transformer models get into kind of Transformer models get into kind of Transformer models get into smaller form factors with less power smaller form factors with less power smaller form factors with less power consumption but consumption but consumption but memory yeah well they're talking about

memory yeah well they're talking about memory yeah well they're talking about ultra um Ultra ultra um Ultra ultra um Ultra bandwidth memory so you bandwidth memory so you bandwidth memory so you ubm not H or yeah so the new memory have ubm not H or yeah so the new memory have ubm not H or yeah so the new memory have to buy not just high bandwidth memory to buy not just high bandwidth memory to buy not just high bandwidth memory Ultra bandwidth memory not to be Ultra bandwidth memory not to be Ultra bandwidth memory not to be confused with super ultra or Mega Ultra confused with super ultra or Mega Ultra confused with super ultra or Mega Ultra yeah it is still an optimization play so yeah it is still an optimization play so yeah it is still an optimization play so until we start to find you a different until we start to find you a different until we start to find you a different way to do the technology for this mod way to do the technology for this mod way to do the technology for this mod because again as you as as said it's because again as you as as said it's because again as you as as said it's super Computing on a buch of neural super Computing on a buch of neural super Computing on a buch of neural network which is just basic simple math network which is just basic simple math network which is just basic simple math we just zillions of them so you can we just zillions of them so you can we just zillions of them so you can optimize that as much as you can with optimize that as much as you can with optimize that as much as you can with but you're not going to reduce the but you're not going to reduce the but you're not going to reduce the amount of memory it uses you're going to amount of memory it uses you're going to amount of memory it uses you're going to have faster memory so you go between have faster memory so you go between have faster memory so you go between your disc and your memory quicker so your disc and your memory quicker so your disc and your memory quicker so it's optimization play and again nothing it's optimization play and again nothing it's optimization play and again nothing wrong in optimization you can do be very wrong in optimization you can do be very wrong in optimization you can do be very creative and and put a lot of innovation creative and and put a lot of innovation creative and and put a lot of innovation in optimization but it's not a in optimization but it's not a in optimization but it's not a fundamental change in the way it works fundamental change in the way it works fundamental change in the way it works so no it it isn't but you know we were so no it it isn't but you know we were so no it it isn't but you know we were talking about earlier um you know analog talking about earlier um you know analog talking about earlier um you know analog and your morph those those suggest a and your morph those those suggest a and your morph those those suggest a different way of approaching that could different way of approaching that could different way of approaching that could be different problem and maybe a be different problem and maybe a be different problem and maybe a slightly AR different architecture but slightly AR different architecture but slightly AR different architecture but there was a cool session his name was um there was a cool session his name was um there was a cool session his name was um Paul Masters he's a CTO of a company Paul Masters he's a CTO of a company Paul Masters he's a CTO of a company called called called um um core um corami so core

um um core um corami so core um um core um corami so core tsunami um he was suggesting that we tsunami um he was suggesting that we tsunami um he was suggesting that we have to think outside of the box you have to think outside of the box you have to think outside of the box you know uh get beyond the Von Newman B yeah architectures yeah and then um B yeah architectures yeah and then um simplifying you know just adapting to simplifying you know just adapting to simplifying you know just adapting to the needs of this next generation of the needs of this next generation of the needs of this next generation of computing or AIU acceler what you call computing or AIU acceler what you call computing or AIU acceler what you call it so um it's interesting because he's it so um it's interesting because he's it so um it's interesting because he's done a couple of things they they've done a couple of things they they've done a couple of things they they've created some as6 as well as some um it's created some as6 as well as some um it's created some as6 as well as some um it's called Fine called Fine called Fine teamed algorithms for things like um teamed algorithms for things like um teamed algorithms for things like um zero you call it zero knowledge proof zero you call it zero knowledge proof zero you call it zero knowledge proof right so you crypto right so you crypto right so you crypto folks uh you know you probably folks uh you know you probably folks uh you know you probably appreciate this uh also he's done some appreciate this uh also he's done some appreciate this uh also he's done some stuff for uh homomorphic security stuff for uh homomorphic security stuff for uh homomorphic security encryption uh which is you know I mean encryption uh which is you know I mean encryption uh which is you know I mean that's a massive overhead right so they that's a massive overhead right so they that's a massive overhead right so they have a um they have an Asic and have a um they have an Asic and have a um they have an Asic and algorithm that you know dramtically algorithm that you know dramtically algorithm that you know dramtically supposedly grammatically accelerate supposedly grammatically accelerate supposedly grammatically accelerate those types of those types of those types of workflows um and so you know the big workflows um and so you know the big workflows um and so you know the big question is why not why not question is why not why not question is why not why not Transformer right having as for a Transformer right having as for a Transformer right having as for a Transformer like we're talking about but Transformer like we're talking about but Transformer like we're talking about but specifically tuning for a particular specifically tuning for a particular specifically tuning for a particular kind of algorithm slash model and um and

kind of algorithm slash model and um and kind of algorithm slash model and um and and Transformer architecture so yeah um and Transformer architecture so yeah um and Transformer architecture so yeah um I think we'll see some of that stuff I think we'll see some of that stuff I think we'll see some of that stuff like I was um there's a company called like I was um there's a company called like I was um there's a company called Green waves that has a gap it's called Green waves that has a gap it's called Green waves that has a gap it's called The Gap n it's a risk five based audio The Gap n it's a risk five based audio The Gap n it's a risk five based audio processor with AI and they were showing processor with AI and they were showing processor with AI and they were showing how they could kind of you know it's how they could kind of you know it's how they could kind of you know it's pretty cool demo it was like you know pretty cool demo it was like you know pretty cool demo it was like you know play a Spotify song and it automatically play a Spotify song and it automatically play a Spotify song and it automatically strips out in real time all of the strips out in real time all of the strips out in real time all of the vocals uh from the so it's pretty pretty vocals uh from the so it's pretty pretty vocals uh from the so it's pretty pretty amazing horsepower for like something amazing horsepower for like something amazing horsepower for like something that runs on a mowatt uh but these are that runs on a mowatt uh but these are that runs on a mowatt uh but these are that's for like herbl and stuff so that that's for like herbl and stuff so that that's for like herbl and stuff so that that's kind of their Market is mostly that's kind of their Market is mostly that's kind of their Market is mostly herbl and things but you're going to see herbl and things but you're going to see herbl and things but you're going to see a lot more I think very functionally f a lot more I think very functionally f a lot more I think very functionally f focused as6 that are designed around AI focused as6 that are designed around AI focused as6 that are designed around AI algorithms and models and you're like algorithms and models and you're like algorithms and models and you're like you know I am accelerating this model you know I am accelerating this model you know I am accelerating this model this Asic accelerates this model purpose this Asic accelerates this model purpose this Asic accelerates this model purpose and um then you really get some and um then you really get some and um then you really get some optimization there on cost and and power optimization there on cost and and power optimization there on cost and and power and things like that so yeah and and I and things like that so yeah and and I and things like that so yeah and and I think that's where definitely Nvidia think that's where definitely Nvidia think that's where definitely Nvidia faces a lot of risk and it can happen faces a lot of risk and it can happen faces a lot of risk and it can happen really quick right I mean going back to really quick right I mean going back to really quick right I mean going back to if the problem is big enough the Asic is if the problem is big enough the Asic is if the problem is big enough the Asic is g to be your answer and if Asic is your g to be your answer and if Asic is your g to be your answer and if Asic is your answer answer answer um and um you don't have to solve your um and um you don't have to solve your um and um you don't have to solve your big problems using high margin highly big problems using high margin highly big problems using high margin highly expensive uh GPU or new equipment too is expensive uh GPU or new equipment too is expensive uh GPU or new equipment too is like I've heard a lot of folks say I'd like I've heard a lot of folks say I'd like I've heard a lot of folks say I'd like to run I'd like to solve this like to run I'd like to solve this like to run I'd like to solve this problem with the equipment that I have problem with the equipment that I have problem with the equipment that I have that may be yeah Intel Nooks or whatever

that may be yeah Intel Nooks or whatever that may be yeah Intel Nooks or whatever they have and they're like I'd rather they have and they're like I'd rather they have and they're like I'd rather just and I just solv the problem with just and I just solv the problem with just and I just solv the problem with the with the with the stuff I already the with the with the stuff I already the with the with the stuff I already depreciated from last year instead of depreciated from last year instead of depreciated from last year instead of buying another thing yeah I know really buying another thing yeah I know really buying another thing yeah I know really want to do that somehow and here's the want to do that somehow and here's the want to do that somehow and here's the crazy thing one of the Keynotes they're crazy thing one of the Keynotes they're crazy thing one of the Keynotes they're talking about how a lot of the super oh talking about how a lot of the super oh talking about how a lot of the super oh yeah it's the supercomputing talk a lot yeah it's the supercomputing talk a lot yeah it's the supercomputing talk a lot of these uh institutes that run these of these uh institutes that run these of these uh institutes that run these large um supercomputers are looking at large um supercomputers are looking at large um supercomputers are looking at extending the life of the supercomputers extending the life of the supercomputers extending the life of the supercomputers not drawing them in but that contrasts not drawing them in but that contrasts not drawing them in but that contrasts with what you know Nvidia and now most with what you know Nvidia and now most with what you know Nvidia and now most recent AMD announced where they want to recent AMD announced where they want to recent AMD announced where they want to have a one-year Cadence on the hardware have a one-year Cadence on the hardware have a one-year Cadence on the hardware and the system it's like wait a minute and the system it's like wait a minute and the system it's like wait a minute that doesn't jive so there's going to be that doesn't jive so there's going to be that doesn't jive so there's going to be this like this like this like really come to Jesus moment well yeah really come to Jesus moment well yeah really come to Jesus moment well yeah you can't have you can't have a yearly you can't have you can't have a yearly you can't have you can't have a yearly supply side Cadence and a threeyear buy supply side Cadence and a threeyear buy supply side Cadence and a threeyear buy side you know Cadence that doesn't seven side you know Cadence that doesn't seven side you know Cadence that doesn't seven years seven years seven years so they're years seven years seven years so they're years seven years seven years so they're going you know hyperscalers hypers you going you know hyperscalers hypers you going you know hyperscalers hypers you know that's kind of like the direction know that's kind of like the direction know that's kind of like the direction hyperscalers are going hyperscalers are going hyperscalers are going but um it but they're not really super but um it but they're not really super but um it but they're not really super Computing right um their stuff Computing right um their stuff Computing right um their stuff depreciates a lot faster with super depreciates a lot faster with super depreciates a lot faster with super Computing there's just I mean there be Computing there's just I mean there be Computing there's just I mean there be thepoke jobs right I mean thepoke jobs right I mean thepoke jobs right I mean there not everyone does are freaking there not everyone does are freaking there not everyone does are freaking global weather global weather global weather model you know but there's probably

model you know but there's probably model you know but there's probably enough demand now to fuel that but as enough demand now to fuel that but as enough demand now to fuel that but as soon as we reach a certain level because soon as we reach a certain level because soon as we reach a certain level because there's not enough of these ships being there's not enough of these ships being there's not enough of these ships being sold or shipped from what I understand sold or shipped from what I understand sold or shipped from what I understand to the demand of AI but that will that to the demand of AI but that will that to the demand of AI but that will that that will slow down at some point I mean that will slow down at some point I mean that will slow down at some point I mean I think Microsoft said there spending 14 I think Microsoft said there spending 14 I think Microsoft said there spending 14 billion capex this year on data center billion capex this year on data center billion capex this year on data center buildout so exactly so there's a there's buildout so exactly so there's a there's buildout so exactly so there's a there's a few there's a probably few good years a few there's a probably few good years a few there's a probably few good years for the Nvidia and other but at some for the Nvidia and other but at some for the Nvidia and other but at some point yeah he will say oh come point yeah he will say oh come point yeah he will say oh come on yeah and you know there's there's on yeah and you know there's there's on yeah and you know there's there's also some other other also some other other also some other other narrative narrative dysfunction narrative narrative dysfunction narrative narrative dysfunction happening here like for instance there's happening here like for instance there's happening here like for instance there's a lot of there's a lot of an anxiety a lot of there's a lot of an anxiety a lot of there's a lot of an anxiety well here's one um so everyone's talking well here's one um so everyone's talking well here's one um so everyone's talking about oh my God it's going to be an about oh my God it's going to be an about oh my God it's going to be an energy crisis we're going to need more F energy crisis we're going to need more F energy crisis we're going to need more F to meet all of the demands that the to meet all of the demands that the to meet all of the demands that the compute that work anticipate somebody is compute that work anticipate somebody is compute that work anticipate somebody is for fact going to be for fact going to be for fact going to be required SED an answer to this yet on required SED an answer to this yet on required SED an answer to this yet on the flip side on the side nobody can the flip side on the side nobody can the flip side on the side nobody can explain who's gonna take for this and explain who's gonna take for this and explain who's gonna take for this and what the application is it's just you what the application is it's just you what the application is it's just you know when you do listen to the Corpus of know when you do listen to the Corpus of know when you do listen to the Corpus of what people are estimating this huge new what people are estimating this huge new what people are estimating this huge new driver of expension for sem industry driver of expension for sem industry driver of expension for sem industry okay it's um super okay it's um super okay it's um super Computing do you know what I'm saying Computing do you know what I'm saying Computing do you know what I'm saying it's like okay if that's superc it's like okay if that's superc it's like okay if that's superc Computing that's going to help the

inv EOS system inv EOS system and and and yeah then you know um if it's that then yeah then you know um if it's that then yeah then you know um if it's that then what what problems are you solving this what what problems are you solving this what what problems are you solving this your training models and how much your training models and how much your training models and how much inflence is there really going to be inflence is there really going to be inflence is there really going to be when I mean when I mean when I mean truly the the inference is going to have truly the the inference is going to have truly the the inference is going to have to be distributed right because no one's to be distributed right because no one's to be distributed right because no one's going to be able to afford to subsidize going to be able to afford to subsidize going to be able to afford to subsidize and pay for and make margin on anything and pay for and make margin on anything and pay for and make margin on anything that's done in the cloud right that's done in the cloud right that's done in the cloud right yeah it'll be expensive so it's just yeah it'll be expensive so it's just yeah it'll be expensive so it's just going to be more expensive if like you going to be more expensive if like you going to be more expensive if like you were saying Pete if you have all this were saying Pete if you have all this were saying Pete if you have all this you know like Headroom that you have you know like Headroom that you have you know like Headroom that you have underutilized Edge or endpoint devices underutilized Edge or endpoint devices underutilized Edge or endpoint devices why don't you just use that stuff and why don't you just use that stuff and why don't you just use that stuff and one of the things Apple's improving is one of the things Apple's improving is one of the things Apple's improving is you don't need 40 tops to do a lot of you don't need 40 tops to do a lot of you don't need 40 tops to do a lot of cool AI stuff right you can do it with cool AI stuff right you can do it with cool AI stuff right you can do it with 11 yeah or less well like I said it's 11 yeah or less well like I said it's 11 yeah or less well like I said it's sometimes it's more RAM issues than top sometimes it's more RAM issues than top sometimes it's more RAM issues than top issues right right but then you know issues right right but then you know issues right right but then you know Apple Apple devices tend to have lower Apple Apple devices tend to have lower Apple Apple devices tend to have lower RAM as well it's just they T they're RAM as well it's just they T they're RAM as well it's just they T they're more they're tightly integrated right more they're tightly integrated right more they're tightly integrated right yeah than Android devices yeah I'm yeah than Android devices yeah I'm yeah than Android devices yeah I'm seeing some interesting stuff people are seeing some interesting stuff people are seeing some interesting stuff people are doing like very focused rag based um doing like very focused rag based um doing like very focused rag based um kind of uh small language models for kind of uh small language models for kind of uh small language models for very focused scenarios so like you train very focused scenarios so like you train very focused scenarios so like you train it on a very focused chunk of data we it on a very focused chunk of data we it on a very focused chunk of data we actually had the student train create a

actually had the student train create a actually had the student train create a tiny ml chatbot trained on all of our tiny ml chatbot trained on all of our tiny ml chatbot trained on all of our YouTube videos actually I was playing YouTube videos actually I was playing YouTube videos actually I was playing with it this week it was pretty cool so with it this week it was pretty cool so with it this week it was pretty cool so if you could ask it all kinds of if you could ask it all kinds of if you could ask it all kinds of questions about tiny ML and questions about tiny ML and questions about tiny ML and state-ofthe-art and it would it would state-ofthe-art and it would it would state-ofthe-art and it would it would cough up all kinds of cool answers based cough up all kinds of cool answers based cough up all kinds of cool answers based on all the because we have hundreds and on all the because we have hundreds and on all the because we have hundreds and thousands of papers and videos and thousands of papers and videos and thousands of papers and videos and things but there was one scenario where things but there was one scenario where things but there was one scenario where where uh someone was talking about they where uh someone was talking about they where uh someone was talking about they make Hospital sensors so the sensors are make Hospital sensors so the sensors are make Hospital sensors so the sensors are all around the hospital room yeah you all around the hospital room yeah you all around the hospital room yeah you can imagine there's like a hundred can imagine there's like a hundred can imagine there's like a hundred different things being measured in a different things being measured in a different things being measured in a patient's room yeah like you could have patient's room yeah like you could have patient's room yeah like you could have a large language model actually a large language model actually a large language model actually interpret all that sensor data so when interpret all that sensor data so when interpret all that sensor data so when the nurse walks in in the morning they the nurse walks in in the morning they the nurse walks in in the morning they could say how's the patient been doing could say how's the patient been doing could say how's the patient been doing you know since last night and they can you know since last night and they can you know since last night and they can get like a natural language explanation get like a natural language explanation get like a natural language explanation of all this data right rather than of all this data right rather than of all this data right rather than reading through a 100 different sensors reading through a 100 different sensors reading through a 100 different sensors themselves which you know so things like themselves which you know so things like themselves which you know so things like that where you're using generative AI to that where you're using generative AI to that where you're using generative AI to sort of take a lot of complex sensor sort of take a lot of complex sensor sort of take a lot of complex sensor data and make sense out of it you know data and make sense out of it you know data and make sense out of it you know in a very focused way can be super in a very focused way can be super in a very focused way can be super valuable or or a car mechanic where valuable or or a car mechanic where valuable or or a car mechanic where where he's got a tool rather than just where he's got a tool rather than just where he's got a tool rather than just giving me obb Port codes like you know giving me obb Port codes like you know giving me obb Port codes like you know tell me what's wrong with this engine tell me what's wrong with this engine tell me what's wrong with this engine and this llm can say okay it's trained and this llm can say okay it's trained and this llm can say okay it's trained on the BMW sensor data and it can say on the BMW sensor data and it can say on the BMW sensor data and it can say you know what I think it's really this you know what I think it's really this you know what I think it's really this so you might want to check this first so so you might want to check this first so so you might want to check this first so I think we're going to see llms used in I think we're going to see llms used in I think we're going to see llms used in these very focused ways running on Edge these very focused ways running on Edge these very focused ways running on Edge Equipment trained on very focused Equipment trained on very focused Equipment trained on very focused corpuses of data and I think that's corpuses of data and I think that's corpuses of data and I think that's going to be super valuable because going to be super valuable because going to be super valuable because that's going to really be at the point that's going to really be at the point that's going to really be at the point of impact in solving that particular of impact in solving that particular of impact in solving that particular problem it's not scarlet youran and chat problem it's not scarlet youran and chat problem it's not scarlet youran and chat Bots pretending to be girlfriend or Bots pretending to be girlfriend or Bots pretending to be girlfriend or whatever no that that's not that's not

whatever no that that's not that's not whatever no that that's not that's not really where it's at these days it's really where it's at these days it's really where it's at these days it's it's about solving real problems and it's about solving real problems and it's about solving real problems and you you can also Imagine Solutions where you you can also Imagine Solutions where you you can also Imagine Solutions where actually you have a kind of a generalist actually you have a kind of a generalist actually you have a kind of a generalist llm that drives you to the specialized llm that drives you to the specialized llm that drives you to the specialized LM you're just G to have to teach those LM you're just G to have to teach those LM you're just G to have to teach those LMS to actually realize that I don't LMS to actually realize that I don't LMS to actually realize that I don't know about this subject let me ask the know about this subject let me ask the know about this subject let me ask the other guy the day I have doing making up other guy the day I have doing making up other guy the day I have doing making up an answer yeah exactly so the actually I an answer yeah exactly so the actually I an answer yeah exactly so the actually I think it's an interesting point the day think it's an interesting point the day think it's an interesting point the day you start seeing that I think you start you start seeing that I think you start you start seeing that I think you start to see what I would consider to see what I would consider to see what I would consider intelligence I mean spe can tell me I intelligence I mean spe can tell me I intelligence I mean spe can tell me I don't know self awareness is very don't know self awareness is very don't know self awareness is very important to go ask Cloe version X you important to go ask Cloe version X you important to go ask Cloe version X you know that's true let me hand this let me know that's true let me hand this let me know that's true let me hand this let me hand this job off to uh somebody who hand this job off to uh somebody who hand this job off to uh somebody who knows better then we started to have knows better then we started to have knows better then we started to have something so but interestingly enough something so but interestingly enough something so but interestingly enough it's the same thing as what you were it's the same thing as what you were it's the same thing as what you were saying about about tinml and also saying about about tinml and also saying about about tinml and also specialized as is we're going to go back specialized as is we're going to go back specialized as is we're going to go back to system integrator job so probably the to system integrator job so probably the to system integrator job so probably the best thing you have to do to be a system best thing you have to do to be a system best thing you have to do to be a system integrator yeah because there integrator yeah because there integrator yeah because there specialization in componentization more specialization in componentization more specialization in componentization more and more so and you need to be to to and more so and you need to be to to and more so and you need to be to to know all about all these specialities know all about all these specialities know all about all these specialities and being capable of assembling them and being capable of assembling them and being capable of assembling them into something into something into something that so yeah I um I that so yeah I um I that so yeah I um I think the D the data problem is just think the D the data problem is just think the D the data problem is just going to be a persistent one oh and and going to be a persistent one oh and and going to be a persistent one oh and and you know thing is you know thing is you know thing is is they already have something I mean in

is they already have something I mean in is they already have something I mean in an adversarial way that's what a model an adversarial way that's what a model an adversarial way that's what a model or a mixture expert is or a mixture expert is or a mixture expert is um you know you have a a collective of um you know you have a a collective of um you know you have a a collective of models some more fine tuned and expert models some more fine tuned and expert models some more fine tuned and expert in certain domains and they they kind of in certain domains and they they kind of in certain domains and they they kind of come up with the better come up with the better come up with the better answer but I I mean that's what's gotten answer but I I mean that's what's gotten answer but I I mean that's what's gotten GPT 4 GPT 4 GPT 4 closer closer closer to or improve the performance of it but to or improve the performance of it but to or improve the performance of it but this whole idea this whole idea this whole idea to scale these models to a societal to scale these models to a societal to scale these models to a societal level I think you just get the problems level I think you just get the problems level I think you just get the problems it becomes a mirror of society right I it becomes a mirror of society right I it becomes a mirror of society right I think it's one thing is yeah um you know think it's one thing is yeah um you know think it's one thing is yeah um you know one head one head one head or uh two heads are better than one type or uh two heads are better than one type or uh two heads are better than one type of theory of this um expanding the of theory of this um expanding the of theory of this um expanding the knowledge base and expertise and the knowledge base and expertise and the knowledge base and expertise and the probability that through AR argument and probability that through AR argument and probability that through AR argument and debate you can get to a better Insight debate you can get to a better Insight debate you can get to a better Insight or result I don't know how far that can get result I don't know how far that can get expanded out but that's why these guys expanded out but that's why these guys expanded out but that's why these guys are trying to scale out now these are trying to scale out now these are trying to scale out now these mixture of experts to like some mixture of experts to like some mixture of experts to like some ridiculous scale right but does that ridiculous scale right but does that ridiculous scale right but does that really solve the problem hallucination really solve the problem hallucination really solve the problem hallucination and everything else to your point the and everything else to your point the and everything else to your point the the fundamental of the technology is the fundamental of the technology is the fundamental of the technology is that if the data is not created well that if the data is not created well that if the data is not created well you're not going to get something that you're not going to get something that you're not going to get something that make any sense so there no there's no

make any sense so there no there's no make any sense so there no there's no again there's no sign right now no no again there's no sign right now no no again there's no sign right now no no evidence that there is a technology that evidence that there is a technology that evidence that there is a technology that can make actually this separation can make actually this separation can make actually this separation between oh this is an argument versus between oh this is an argument versus between oh this is an argument versus this is a fact right well you have to this is a fact right well you have to this is a fact right well you have to have the the DAT sets have to be clean have the the DAT sets have to be clean have the the DAT sets have to be clean the DAT large language from a large the DAT large language from a large the DAT large language from a large language model there's no difference language model there's no difference language model there's no difference between a sentence that is a fact and a between a sentence that is a fact and a between a sentence that is a fact and a sentence that is is just a bunch of word sentence that is is just a bunch of word sentence that is is just a bunch of word which statistically we going there's which statistically we going there's which statistically we going there's more arguments you're going to have more more arguments you're going to have more more arguments you're going to have more the answer the fact so again you have to the answer the fact so again you have to the answer the fact so again you have to keep the fundamentals in mind I'm not keep the fundamentals in mind I'm not keep the fundamentals in mind I'm not saying that those tools can be used for saying that those tools can be used for saying that those tools can be used for very interesting and Powerful things but very interesting and Powerful things but very interesting and Powerful things but you have to keep in mind the you have to keep in mind the you have to keep in mind the fundamentals otherwise yeah itations fundamentals otherwise yeah itations fundamentals otherwise yeah itations yeah but really quick I mean Pete I yeah but really quick I mean Pete I yeah but really quick I mean Pete I think you're on this a lot of good stuff think you're on this a lot of good stuff think you're on this a lot of good stuff in you know using uh llms and vision in you know using uh llms and vision in you know using uh llms and vision models to index things or to classify models to index things or to classify models to index things or to classify things and then maybe things and then maybe things and then maybe um create embeddings right um but then um create embeddings right um but then um create embeddings right um but then again all those mechanisms don't tend to again all those mechanisms don't tend to again all those mechanisms don't tend to be exact they're probabilistic be exact they're probabilistic be exact they're probabilistic unfortunately and um but I mean there unfortunately and um but I mean there unfortunately and um but I mean there those are like some some of those use those are like some some of those use those are like some some of those use cases that like you said that's where cases that like you said that's where cases that like you said that's where you're going to realize value it's you're going to realize value it's you're going to realize value it's nothing sexy right but if you can nothing sexy right but if you can nothing sexy right but if you can channel that into solving specific channel that into solving specific channel that into solving specific problems in an overall system problems in an overall system problems in an overall system yeah but is it going to transform the yeah but is it going to transform the yeah but is it going to transform the world probably not but it's gonna it's world probably not but it's gonna it's world probably not but it's gonna it's going to improve some stuff

going to improve some stuff going to improve some stuff classification and prediction which are classification and prediction which are classification and prediction which are the two basic things that we do with the two basic things that we do with the two basic things that we do with these models is extremely effective in a these models is extremely effective in a these models is extremely effective in a number of domains and it can I think number of domains and it can I think number of domains and it can I think that you know the interesting thing was that you know the interesting thing was that you know the interesting thing was a I was thinking about that you know a I was thinking about that you know a I was thinking about that you know recently is a we seems to have to be recently is a we seems to have to be recently is a we seems to have to be kind of have this quest to uh to replace kind of have this quest to uh to replace kind of have this quest to uh to replace human intelligence where we we haven't human intelligence where we we haven't human intelligence where we we haven't actually solved some of the basic actually solved some of the basic actually solved some of the basic problems that are labor intensive or or problems that are labor intensive or or problems that are labor intensive or or time intensive humans lack time intensive humans lack time intensive humans lack classification lack prediction and that classification lack prediction and that classification lack prediction and that could enormous help if you if you scale could enormous help if you if you scale could enormous help if you if you scale that and if you make it better and that and if you make it better and that and if you make it better and better and better as opposed to trying better and better as opposed to trying better and better as opposed to trying to have something that talks to you and to have something that talks to you and to have something that talks to you and look like you're human and will tell you look like you're human and will tell you look like you're human and will tell you the same stupid things that a human does the same stupid things that a human does the same stupid things that a human does so right well at a broader at a broader so right well at a broader at a broader so right well at a broader at a broader Society level I wish we put more as much Society level I wish we put more as much Society level I wish we put more as much energy into educating students as we did energy into educating students as we did energy into educating students as we did creating AI but that's a whole other oh creating AI but that's a whole other oh creating AI but that's a whole other oh yeah we should we should have a we yeah we should we should have a we yeah we should we should have a we should have a whole episode on that we should have a whole episode on that we should have a whole episode on that we should have dude I know but should have dude I know but should have dude I know but that's in this sense I do believe that that's in this sense I do believe that that's in this sense I do believe that we are but but but especially in the US we are but but but especially in the US we are but but but especially in the US and actually in more and more countries and actually in more and more countries and actually in more and more countries that nobody's investing in education that nobody's investing in education that nobody's investing in education anymore but this is where we should anymore but this is where we should anymore but this is where we should leverage llms but but this we should we leverage llms but but this we should we leverage llms but but this we should we should completely change the way we should completely change the way we should completely change the way we teach to people fire all the fire teach to people fire all the fire teach to people fire all the fire everybody and restart from scratch everybody and restart from scratch everybody and restart from scratch because this is a because this is a because this is a I don't know about that you almost dial I don't know about that you almost dial I don't know about that you almost dial it down well you know it's Demetri what it down well you know it's Demetri what it down well you know it's Demetri what else can you else can you else can you expect everyone start over no but guys

expect everyone start over no but guys expect everyone start over no but guys if you look at the human race over if if you look at the human race over if if you look at the human race over if you look at the RO statistics in France you look at the RO statistics in France you look at the RO statistics in France there's there's close to a million there's there's close to a million there's there's close to a million teachers I think 850 million it's on teachers I think 850 million it's on teachers I think 850 million it's on average 15 per student I mean 15 student average 15 per student I mean 15 student average 15 per student I mean 15 student per per per per teacher okay and if you per per per per teacher okay and if you per per per per teacher okay and if you go into the CL yeah you are just the ra go into the CL yeah you are just the ra go into the CL yeah you are just the ra number but in the classroom there's number but in the classroom there's number but in the classroom there's always 40 guys there's always 40 always 40 guys there's always 40 always 40 guys there's always 40 students and one teacher so what's going students and one teacher so what's going students and one teacher so what's going on and that's very human intensive labor on and that's very human intensive labor on and that's very human intensive labor and what I'm saying is that yeah a and what I'm saying is that yeah a and what I'm saying is that yeah a number of basic knowledge teaching you number of basic knowledge teaching you number of basic knowledge teaching you can replace that by machines and then can replace that by machines and then can replace that by machines and then use the human for the critical thinking use the human for the critical thinking use the human for the critical thinking aspect I mean how much of the time you aspect I mean how much of the time you aspect I mean how much of the time you spend doing critical thinking versus spend doing critical thinking versus spend doing critical thinking versus learning raw knowledge in class no learning raw knowledge in class no learning raw knowledge in class no that's true the raw knowled is probably that's true the raw knowled is probably that's true the raw knowled is probably 80 I mean it's probably 90 95% of the 80 I mean it's probably 90 95% of the 80 I mean it's probably 90 95% of the time you can you can replace that by time you can you can replace that by time you can you can replace that by machine machine machine and then have like 20% or 40% of the and then have like 20% or 40% of the and then have like 20% or 40% of the time to do critical thinking and still time to do critical thinking and still time to do critical thinking and still do a better job so yeah I mean teachers do a better job so yeah I mean teachers do a better job so yeah I mean teachers do teachers do a lot more than just do teachers do a lot more than just do teachers do a lot more than just teaching wrote information right they're teaching wrote information right they're teaching wrote information right they're they're they're there's a lot of you they're they're there's a lot of you they're they're there's a lot of you know kind of support about how to learn know kind of support about how to learn know kind of support about how to learn should but I think their job should be should but I think their job should be should but I think their job should be focused more more on the critical focused more more on the critical focused more more on the critical thinking and it should be a tool for thinking and it should be a tool for thinking and it should be a tool for them to because I mean they are boring them to because I mean they are boring them to because I mean they are boring teachers I had even at high level a math teachers I had even at high level a math teachers I had even at high level a math teacher who was just teaching course the teacher who was just teaching course the teacher who was just teaching course the jokes were in the margin so we knew the jokes were in the margin so we knew the jokes were in the margin so we knew the jokes from the previous year and this jokes from the previous year and this jokes from the previous year and this guy were just he was telling his stuff guy were just he was telling his stuff guy were just he was telling his stuff sure yeah I mean that's could have been sure yeah I mean that's could have been sure yeah I mean that's could have been 100% replaced by a machine 100% sure

100% replaced by a machine 100% sure 100% replaced by a machine 100% sure yeah but you know we also have to yeah but you know we also have to yeah but you know we also have to remember yeah there there the babysitter remember yeah there there the babysitter remember yeah there there the babysitter kids too people forget that oh hey kids too people forget that oh hey kids too people forget that oh hey that's a good idea that's a great idea that's a good idea that's a great idea that's a good idea that's a great idea what ANM to babysit what ANM to babysit what ANM to babysit kids no no seriously this is a great kids no no seriously this is a great kids no no seriously this is a great idea yeah this is a great idea why don't idea yeah this is a great idea why don't idea yeah this is a great idea why don't you work on that you know I don't know you work on that you know I don't know you work on that you know I don't know if I should find that insulting or not I if I should find that insulting or not I if I should find that insulting or not I mean I I put out like the dumbest idea mean I I put out like the dumbest idea mean I I put out like the dumbest idea and he thinks it's brilliant go is there and he thinks it's brilliant go is there and he thinks it's brilliant go is there something wrong with that I don't know something wrong with that I don't know something wrong with that I don't know but look at the look at the number of but look at the look at the number of but look at the look at the number of stupid things that people buy I'm sorry stupid things that people buy I'm sorry stupid things that people buy I'm sorry I think and you know what I'm G to I think and you know what I'm G to I think and you know what I'm G to counter argue that I believe that this counter argue that I believe that this counter argue that I believe that this Sal could sometimes do a better job that Sal could sometimes do a better job that Sal could sometimes do a better job that some babysit on some parents and some some babysit on some parents and some some babysit on some parents and some parents oh yeah parents oh yeah parents oh yeah yeah so yeah so yeah so there you go humanoid Rob Next Gen there you go humanoid Rob Next Gen there you go humanoid Rob Next Gen Keynote I think Keynote I think Keynote I think forc yeah generative ey babysitter forc yeah generative ey babysitter forc yeah generative ey babysitter that'll work pretty well can you raise that'll work pretty well can you raise that'll work pretty well can you raise your kids better than my a our our our your kids better than my a our our our your kids better than my a our our our llms can actually raise your other llms can actually raise your other llms can actually raise your other better than you can think what kind of better than you can think what kind of better than you can think what kind of value that I would guess yes for a value that I would guess yes for a value that I would guess yes for a larger number yikes but on the other larger number yikes but on the other larger number yikes but on the other hand we could probably use LMS to teach hand we could probably use LMS to teach hand we could probably use LMS to teach parents to be better parents H yeah parents to be better parents H yeah parents to be better parents H yeah parental coaches yeah you know maybe parental coaches yeah you know maybe parental coaches yeah you know maybe could be yeah I don't know I don't know could be yeah I don't know I don't know could be yeah I don't know I don't know by the way there's lectured by a machine by the way there's lectured by a machine by the way there's lectured by a machine weird idea there's no real such thing as weird idea there's no real such thing as weird idea there's no real such thing as parental coaching I mean in at at scale

parental coaching I mean in at at scale parental coaching I mean in at at scale there used to be when families were there used to be when families were there used to be when families were close together but I mean you you have close together but I mean you you have close together but I mean you you have to go get a license to drive a car you to go get a license to drive a car you to go get a license to drive a car you don't get a license to raise your kid don't get a license to raise your kid don't get a license to raise your kid that's true read the Dr Spock book yeah that's true read the Dr Spock book yeah that's true read the Dr Spock book yeah yeah but you know um it's you know um yeah but you know um it's you know um yeah but you know um it's you know um here here's the thing I think with here here's the thing I think with here here's the thing I think with General General General AI the AI crap AI the AI crap AI the AI crap um who knows they might be closer than um who knows they might be closer than um who knows they might be closer than you know we think because what they're you know we think because what they're you know we think because what they're really trying to do is they're trying to really trying to do is they're trying to really trying to do is they're trying to create a societal create a societal create a societal intelligence that is intelligence that is intelligence that is basically informed by a collection a basically informed by a collection a basically informed by a collection a massive massive massive collection of um you know smaller collection of um you know smaller collection of um you know smaller intelligence that are reflective of intelligence that are reflective of intelligence that are reflective of substan substan substan intelligence and um something to think intelligence and um something to think intelligence and um something to think about you know um and it's a different about you know um and it's a different about you know um and it's a different variety of intelligence that shouldn't variety of intelligence that shouldn't variety of intelligence that shouldn't be compared with human an individual I be compared with human an individual I be compared with human an individual I think an individual has a potential of think an individual has a potential of think an individual has a potential of being much more talented and capable in being much more talented and capable in being much more talented and capable in in domains um in certain specific things in domains um in certain specific things in domains um in certain specific things than an LM ever could or a generative than an LM ever could or a generative than an LM ever could or a generative model ever could model ever could model ever could and the scary thing I think is what what and the scary thing I think is what what and the scary thing I think is what what a collective intelligence an a collective intelligence an a collective intelligence an intelligence that's comprised of a intelligence that's comprised of a intelligence that's comprised of a collection of sub collection of sub collection of sub intelligences um what would that imply

intelligences um what would that imply intelligences um what would that imply you know and what's the end end result you know and what's the end end result you know and what's the end end result of that because then it turn on the TV of that because then it turn on the TV of that because then it turn on the TV you'll know we see yeah just turn on the you'll know we see yeah just turn on the you'll know we see yeah just turn on the TV and and then and then we're just TV and and then and then we're just TV and and then and then we're just going to it's just going to spew out going to it's just going to spew out going to it's just going to spew out stuff that we complain about on a daily stuff that we complain about on a daily stuff that we complain about on a daily basis about Society how broken it is and basis about Society how broken it is and basis about Society how broken it is and how dysfunctional and blah blah blah and how dysfunctional and blah blah blah and how dysfunctional and blah blah blah and and it can only be a mirror of that it and it can only be a mirror of that it and it can only be a mirror of that it can't how can it be more can't how can it be more can't how can it be more Progressive yeah but there's always a Progressive yeah but there's always a Progressive yeah but there's always a there's a how how does the math work out there's a how how does the math work out there's a how how does the math work out somebody has to explain that to me you somebody has to explain that to me you somebody has to explain that to me you can't you can't use a bunch of you know can't you can't use a bunch of you know can't you can't use a bunch of you know techn mumbo jumbo and freaking techn mumbo jumbo and freaking techn mumbo jumbo and freaking Neuroscience Theory to answer that Neuroscience Theory to answer that Neuroscience Theory to answer that question I think it's actually a pretty question I think it's actually a pretty question I think it's actually a pretty simple answer but there is a fundamental simple answer but there is a fundamental simple answer but there is a fundamental problem in this thinking problem in this thinking problem in this thinking as long as we continue to train this as long as we continue to train this as long as we continue to train this model with only human generated data model with only human generated data model with only human generated data yeah it's just a fundamental problem so yeah it's just a fundamental problem so yeah it's just a fundamental problem so unless we had this discussion if we unless we had this discussion if we unless we had this discussion if we start maybe we should start training start maybe we should start training start maybe we should start training models on physics or whatever but if we models on physics or whatever but if we models on physics or whatever but if we have this we always going to have this have this we always going to have this have this we always going to have this bias because the data is human generated bias because the data is human generated bias because the data is human generated so you're only going to get kind of so you're only going to get kind of so you're only going to get kind of approximation of what the human does you approximation of what the human does you approximation of what the human does you might do it quicker faster better not might do it quicker faster better not might do it quicker faster better not better that's better you might do it better that's better you might do it better that's better you might do it quicker and faster but it will still be quicker and faster but it will still be quicker and faster but it will still be it's fundamentally human so the day we it's fundamentally human so the day we it's fundamentally human so the day we start finding ways to feed this model start finding ways to feed this model start finding ways to feed this model with data that is curated or generated with data that is curated or generated with data that is curated or generated by something else than just human maybe by something else than just human maybe by something else than just human maybe there's hope to me yeah well yeah bottom there's hope to me yeah well yeah bottom there's hope to me yeah well yeah bottom line the more useful stuff is definitely line the more useful stuff is definitely line the more useful stuff is definitely showing up on device and yeah you know

showing up on device and yeah you know showing up on device and yeah you know that's highly contextualized channeled that's highly contextualized channeled that's highly contextualized channeled for specific use and tuned yeah tuned for specific use and tuned yeah tuned for specific use and tuned yeah tuned and constrained um based on um you know and constrained um based on um you know and constrained um based on um you know proper risk risk management you know proper risk risk management you know proper risk risk management you know maybe that's what we have to to accept maybe that's what we have to to accept maybe that's what we have to to accept as human we great at making tools and as human we great at making tools and as human we great at making tools and we're trying to replace ourselves that's we're trying to replace ourselves that's we're trying to replace ourselves that's what I don't understand no we're so far what I don't understand no we're so far what I don't understand no we're so far away from that yeah let's keep on making away from that yeah let's keep on making away from that yeah let's keep on making great tools and make sure those tools great tools and make sure those tools great tools and make sure those tools are actually well designed with well are actually well designed with well are actually well designed with well data with human intent human intent and data with human intent human intent and data with human intent human intent and precision um and you know one thing I precision um and you know one thing I precision um and you know one thing I think I'm starting to real I've kind of think I'm starting to real I've kind of think I'm starting to real I've kind of had a hunch of had a hunch of had a hunch of of this but is being increasingly of this but is being increasingly of this but is being increasingly confirmed is generative AI is just confirmed is generative AI is just confirmed is generative AI is just another form of AI has every single another form of AI has every single another form of AI has every single aspect of issues and challenges aspect of issues and challenges aspect of issues and challenges associated with it nothing has changed associated with it nothing has changed associated with it nothing has changed nothing it's nothing it's nothing it's it's just a different it's just a different it's just a different flavor you know of ice cream so pilot flavor you know of ice cream so pilot flavor you know of ice cream so pilot plus plus plus plus yeah but if you want plus plus plus plus yeah but if you want plus plus plus plus yeah but if you want to raise has it own to raise has it own to raise has it own issues everything you have to make issues everything you have to make issues everything you have to make people dream if you want to raise more people dream if you want to raise more people dream if you want to raise more Capital so yeah yeah I know but you know Capital so yeah yeah I know but you know Capital so yeah yeah I know but you know those companies won't be around for very those companies won't be around for very those companies won't be around for very long so you know they long so you know they long so you know they ultimately they feel it in their bones ultimately they feel it in their bones ultimately they feel it in their bones well it's the same humans that generate well it's the same humans that generate well it's the same humans that generate the data that fits the models that make the data that fits the models that make the data that fits the models that make the investment the investment the investment decisions yeah there you go all right

decisions yeah there you go all right decisions yeah there you go all right sounds good should we should we adjourn sounds good should we should we adjourn sounds good should we should we adjourn that was a good session don't you think that was a good session don't you think that was a good session don't you think oh yeah I have to jump on another I have oh yeah I have to jump on another I have oh yeah I have to jump on another I have to jump on another great don't don't to jump on another great don't don't to jump on another great don't don't jump off yet hold on you want to take us jump off yet hold on you want to take us jump off yet hold on you want to take us out who wants to take us out Pete uh out who wants to take us out Pete uh out who wants to take us out Pete uh Pete who that was sure I'll take us out Pete who that was sure I'll take us out Pete who that was sure I'll take us out why not you've been wallowing in more uh why not you've been wallowing in more uh why not you've been wallowing in more uh AI uh wonderings and meanderings for the AI uh wonderings and meanderings for the AI uh wonderings and meanderings for the past hour and a half but uh hopefully past hour and a half but uh hopefully past hour and a half but uh hopefully you've pulled out some good scraps of you've pulled out some good scraps of you've pulled out some good scraps of good uh good Insight with us here and uh good uh good Insight with us here and uh good uh good Insight with us here and uh Leonard holding holding down the fort in Leonard holding holding down the fort in Leonard holding holding down the fort in San Francisco at semicon and Dimitri at San Francisco at semicon and Dimitri at San Francisco at semicon and Dimitri at South Park um we had a good session so South Park um we had a good session so South Park um we had a good session so we're all doing this for not only our we're all doing this for not only our we're all doing this for not only our own entertainment and yours but also for own entertainment and yours but also for own entertainment and yours but also for the kids Elevate our kids so make sure the kids Elevate our kids so make sure the kids Elevate our kids so make sure you go to elevat our kids.org to you go to elevat our kids.org to you go to elevat our kids.org to contribute so that as we were saying contribute so that as we were saying contribute so that as we were saying before the education of kids uh maybe before the education of kids uh maybe before the education of kids uh maybe should be the priority over the creation should be the priority over the creation should be the priority over the creation of artificial intelligence but um you of artificial intelligence but um you of artificial intelligence but um you know we'll see but if you can go there know we'll see but if you can go there know we'll see but if you can go there and donate you'll help bridge the and donate you'll help bridge the and donate you'll help bridge the digital uh divide and um enable people digital uh divide and um enable people digital uh divide and um enable people to have kids to have laptops and other to have kids to have laptops and other to have kids to have laptops and other things that they need to sort of get things that they need to sort of get things that they need to sort of get educated and be be part of society and educated and be be part of society and educated and be be part of society and run those AI run those AI run those AI models they're so important for models they're so important for models they're so important for parenting exactly oh my god um so yeah parenting exactly oh my god um so yeah parenting exactly oh my god um so yeah and like us on YouTube you'll find us on and like us on YouTube you'll find us on and like us on YouTube you'll find us on your favorite podcast and iTunes and your favorite podcast and iTunes and your favorite podcast and iTunes and like share like share like share subscribe click the button whatever

subscribe click the button whatever subscribe click the button whatever button in front of you and uh and yeah button in front of you and uh and yeah button in front of you and uh and yeah and we'll see you next week yeah avoid and we'll see you next week yeah avoid and we'll see you next week yeah avoid the dislike button at all cost oh yeah the dislike button at all cost oh yeah the dislike button at all cost oh yeah thank all right take care guys bye