{
  "video_id": "6mdLXUbcw2Y",
  "channel_slug": "dataindependent",
  "channel_handle": "dataindependent",
  "title": "Siqi Chen - AI Thoughts, Wild Predictions and Musings",
  "duration_seconds": 2021.0,
  "url": "https://www.youtube.com/watch?v=6mdLXUbcw2Y",
  "upload_date": "",
  "transcript": "this is a conversation with siki Chen\nsiki was previously the chief product\nofficer and then CEO of sandbox VR and\nmost recently the CEO of Runway a\nfintech company disrupting the business\nindustry in full transparency I reached\nout to Seiki because of his Twitter\nprofile he has a lot of AI thoughts and\npredictions that I found super\ninteresting in this conversation we\nreviewed living forever unbounded\nintelligence and the place the AI has in\nthe school plays enjoy one of the things\nthat led me to your background was your\nTwitter profile and specifically all the\nAI tweets you're putting out now they're\na mix of philosophy bold predictions and\nreally just cool thoughts that I wanted\nto get your take on so I'm excited to\nhave a conversation with you today me\ntoo thanks for having me absolutely so\nwhat's different about the space that\nwe're seeing right now based off of\nother things that you've seen in your\ncareer\num well with tons of great changes great\nopportunity and this very much reminds\nme of Past Times a great change\nwhen Facebook releases first platform\nthat was very exciting lots of people\nhacking on new products new ideas\nsimilar thing happening with mobile\nuh and we're seeing that same energy now\nand what excites me about things this\ntime around is I think it's going to be\nbigger than mobile and social then\neverything that's come before combined\num and so there's just been never been a\nbetter time to be a builder right now\nyeah absolutely and I was actually I\nwanted to dig into that further What\nanalogy do you use when you're\nexplaining this to people so Mobile's\none but any other analogies you use yeah\nthat's a great question because I think\nuh I have a P with people using\nanalogies to the current change right\npeople compare it to mobile and social\nand electricity like I just did\nuh there's people who compared it to\nnuclear power the atom bomb\nelectricity\nand my peeve is actually I think all of\nthose underrate the magnitude of this\none\num and the way I like to explain it is\nthis is actually the first meta\ntechnology\nwhat I mean by that is if you think\nabout the invention of electricity or\nthe atom bomb or the internet all of\nthose things were byproducts of\nintelligence\nwe've never had the ability to harness\nintelligence itself\nand so that's what makes that such a\nprofound an exciting change and why any\nattempts to draw analogy of what's come\nbefore it's gonna fall short\nsure so we haven't seen the analogy yet\npotentially it's a new thing it's\ndisconnected\nwhere do you find where do you where do\nyou land on the unbounded intelligence\nside of the house the the AI Doomer side\nand it's going to get out of hand how do\nyou think about that\nno\num I am\nI've\nambivalated back and forth I was very\nmuch strong Doomer initially after remix\ngaming hypothesis and thinking about\nthis well and what time what time was\nthis\nthis is about a year and a half ago when\nI first encountered a scaling hypothesis\nfrom Gordon um and that made a case that\nwe are far closer to AGI than most\npeople expect\nthat maybe all it takes is just further\nscaling lots and uh one or two more\nalgorithmic breakthroughs which we can\nget into\nwhere I am now is if you look at the\nlatest advancements in open AI the idea\nof using\nGPT LMS to understand better the war\nKings of lln making it more skewable\nmore understandable has given me a bit\nmore confidence in our ability to steer\nin the right direction\nso I think there is a pretty good case\nto be made that says maybe it'll be okay\na combination of explainability\nunderstandability Regulation\num will help us get there\nthe flip side of that is as capabilities\nincrease as it inevitably will\num what is\ngoing to be the impact of having\neveryone with a laptop wielding the\nequivalent of the power of an animal\npart of say so I do not know I am\ncautiously optimistic I do generally\nthink\nyou know there's an increasing school of\nthought that says this is not going to\nhappen that you know the doomers are all\nlike religious cultists um I love Mark\nAndreessen you know is it a mess from\nthe last two companies I think when I\nhear arguments like that being\npopularized my probability of Doom goes\nup\num and when people are freaking out\nabout him I probably didn't goes down so\nit variables depends yeah I think the\nalmost the room the Romantic part is\nthat nobody knows nobody knows\num and almost by definition it's on it's\nunknowable you brought up open Ai and\nthere well openness some people have\ncriticized in the past saying they\nshould be more more open some have said\nnah they're doing a great job and\nthey're really taking this super slow\nwhere do you think they're landing on\nthe scale what they should be doing\nI think they're making the Right Moves\nobviously there's a reason why they're\nwinning and winning by so much\num from a Doomer perspective it's very\ngood that uh they're closed right um\nwe've seen what even a less capable\nmodel how quickly that can involve in\nthe wild\num but that will also happen too\nI think uh my worry has been okay\nif you\nbelieve that open AI is going to call\nfor regulation regulate itself be\nextremely careful about alignment which\nI think it will and be closed\nwhat also seems to be true is that\nwhat's available out in the world of\nsource is maybe a year to 18 months\nbehind\nand so what happens in a world where\nsomeone with less Scruples than opening\nup\njust\ncrates and spreads towards a place where\nthe L and more the intelligence can\nimprove upon itself\nI think we're within five years of that\nand\ndoes that then overpower a more cautious\norganization\nI'm like my mind trying to tries to rush\nto other mental models or like again\nanalogies to think about well how can I\nuh\nhow can I position my argument to even\nthink about this in the first place but\nit's it's hard to grasp in the in for it\num so then hype Cycles you know you\nstart off super early and then you have\na peak of inflated expectations those\naren't met and then you go down to the\ntrough of like no this isn't as great as\nwe thought necessarily and then it kind\nof evens out to the Middle where do you\nthink we are in the AI hype cycle right\nnow because there's certainly a lot of\nhype yeah it's a mixed bag I mean\nthere's certainly a massive Spike\nin hype in the short term so I would say\nin the short term probably overhyped\nover if you look at a long enough time\nHorizon I would argue it's probably\nunder hyped\nso I think we're\nin a superposition between the two\nthere's still a lot of like skepticism\naround the impact over the long run and\nhow quickly it's going to hit I think\nthat's pretty underrated then you know\nif you look at April it's really after\nthe B4 resulted like oh this is amazing\nit's going to change everything right\nnow\num and I think that's probably overhyped\nwhich I certainly contributed to it it's\na lot of excitement but my excitement\nwas really like not just about the next\nfew months or a few weeks it's like it's\nclose it's here the fact that we're even\ntalking about it is tremendous right\nlike five years ago everything that's\nhappening is pure science fiction\nand we're living in it right now I think\npart of that too is we've only seen the\nfirst order impacts of these language\nmodels like they produce some really\nreally interesting things but we haven't\nyet seen the second order of the\ncompanies building on top of them yet to\ngo do some really cool products and\nproductivity so just starting yeah\nthere's a lot too right both the\nimprovements so there's a second order\neffect and then there's the primary\neffect too of the impact of one people\ngetting more out of gpd4\nthrough better superstructures and uh\nthe models themselves improving and\nbeing capable of self-learning through\nLaura and improving tool usage so\nthere's just a lot there and so you know\nlike what is exponential growth it's\njust linear increase in the rate of\nincrease right and that is what's\nhappening a probably super super\nexponentially today I was on a Twitter\nspace when gbt4 just came out and I\nforget who the person was but they said\nwe haven't even figured out how to use\nGPT 3.5 yet and you just gave us dbt4\nthat's right that's right that's going\nto keep on happening and\num it's only going to keep on going\nquicker as more Frameworks come out as\nwe start to learn these tools better\nthat's right I want to dig into uh a fun\ntopic\nthose Wes Anderson trailers that came\nout so the Matrix in Wes Anderson or\nStar Wars in West Anderson I found\nmyself\nalmost I I couldn't look away\nand they were so fascinating and I and\njust the little details and all the\nsmall little things what do you think it\nwas about those specifically that went\nso viral or why did that capture so much\nattention\num\nyou know it's I think it's fantasy\nfulfillment\nright like I people like Wes Anderson is\nextremely stylized\num it's immediately recognizable\nand\nthe idea that you don't have to just\nlike make a joke oh what if Wes Anderson\nmade a matrix the idea that you know\nwith maybe a few days work and probably\na few seconds work in the coming years\nif you you don't as I have to be a joke\nyou can just make it not only that you\ncan make the trailer it'll make the\nmovie\num I think it captures a sense of\nimagination where people look at a movie\nand you know I wish that um The Matrix\ntwo and three weren't terrible\nand\nI can just ask for that it's almost like\nfan fiction to the max\nright exactly it's Fan Fiction made real\nright and um yeah I think it's both\nvisually evocative but it also\ndemonstrates the potential of what you\ncan do in expressing your own creativity\nand that's going to be something else\nyeah do you think there's a chance we\ncould we as in the kind of culture we\ncould get d sensitized\nto these types of things\num\nthat's really hard to say\num\non the one hand I think on some level\nthat's kind of happening\num\nyou know I think a similar argument was\nmade about Photoshop right you can't\nbelieve pictures anymore\num\nI think\nthere's probably going to be\nthere's a there's a few forces that's\ngoing to make it pretty hard to predict\nthe outcome\nthe first one is\nyou know as the technology improves the\nFidelity and the creativity of it also\nimproves right so you might\nquickly adjust to the idea of like an AI\ncreated Wes Anderson trailer\nbut\nas you get adjusted to that I think the\ntechnology is going to prove faster than\nthat so it's going to be more funny more\ncreative higher Fidelity faster\nwill you adjust to that maybe\nso I think that's one part of this\nI think the other part of this too is as\nthis evolves you may get adjusted to the\nAI creative creative things perhaps even\nif it's good but then you will the value\nof what it what is authentic increases\nso I think that's like from an\ninvestment standpoint what I'm thinking\nabout like in a terminal AI world what\nis valuable\nand things that have like provable\nauthentic provenance probably becomes\nmore valuable\nlike a Chef cooking your food\nlike a handmade movie right the\nauthentic becomes more valuable but that\ndoesn't mean that you're not going to\nfind the other things fun because of the\nshared quality of what can be generated\nthat's technology improves so it's just\ngoing to be purely more fun for everyone\ninvolved it's probably my expectation\nsure and that sounds like a good outcome\nalmost no matter what but\num let me ask how do you define\nauthentic what because the two examples\nyou just gave were almost physical\noccurrences\nwhat is authenticity in the digital\nworld or how do you think about it\num so there's this short story called\nthe metamorphosis of prime intellect\num they talked about the authentic\num and authentic is like that which a\nhuman being without the mediation of\nmachine intelligent technology creates\nby hand\nso a book that is provenly written by a\nperson right like perhaps like there is\na video live stream and someone just\nlike writing a word by word right\num a movie production\num a really good example is actually\num the uh Carl Christopher Nolan uses\nspecial effects right there is a level\nof digital effects but by and large\nthere is an increasing preference to do\npractical effects in movie making\num people just find it\nmore attractive and killing and I think\nthere's going to be a similar movement\nthere\num it is like\nuh the output of the butlerian Jihad\nwhich is a reference to Doom right\num the butler rehab basically in a doom\nWorld basically there was for people who\ndon't know there was a revolution\nagainst um any kind of machine\nintelligence\nand that is like the study of the new\nworld which is very science fiction but\nthere are like no real computers and\neverything there is arguably authentic\num and so those are the things that I\nthink I would Define as it authentic\nand I assume there's also a level of\ntrust with it too because you brought up\nthe part about having a writer the only\nreason why you need a video camera to\nwatch him write the book is it or him or\nher is if you don't believe in the first\nplace right and so I think that the\naspect of human human connection and the\ntrust with that too will go a long way\nas well so you brought up the investment\nhat that you wear as well besides being\nauthentic\nI'm a big fan of elab Gill's list of\nmoats and defensibility and he kind of\ngoes through\num you know your GTM your network\neffects all that good stuff what is top\nof mind for you when it comes to modes\nand defensibility yeah\num\nlike there's only two that I think or\nmaybe three that I think\nour answers I've had to that question\num\nthe first the first one is brand\nis a pretty good moat\nMr Beast has a brand\nand I think that's uh there's a pretty\ngood case to me that Mr Beast is going\nto be one of the most\nfamous and successful valuable people on\nthe in the world\nlike okay it's like you can now talk to\nMr Beast or at least the Incarnation you\nhave a relationship with\nuh all of your fans uh and you know that\na Mr Beast video If created is authentic\nit's him it's not somebody else\num brand is one\nuh my favorite one is probably\nrelationships\num and so a lot of this a lot of my\nthinking comes in thinking through what\nis\nlike what is a company's place in a\nterminal AI world where AGI just exists\nand you can just conjure up anything\nwhat what is a mode in that world\nand I think relationships are a note in\nour relationships I mean\nfor example the relationship that\num Hawking Phoenix has with her in her\nright okay maybe in that world there is\nopen AI that is a superior AI\nbut he's not switching because there's a\nrelationship with that artificial agent\nthat is what's her name\nCharlie Charlie Hanson\nuh so I think that's a note\nthose are probably like two more\nsustainable ones on top of you know the\nusual Network effects and\num customer acquisition but I think\nthose are probably the two of the more\nunderrated ones in the world today when\nwas the last time you watched her with\nwalking Phoenix only watched over once\nit's probably 10 years ago I I need to\nre-watch it with everything that's going\non today and just with new eyes\nit's prescient suppression movie more on\nthe second order effects uh the younger\ngeneration with education children\ninteracting with AI what's your\nprediction or how are you thinking about\nhow kids should be interacting with the\nguy if if at all or anything around that\num I think they should be acting\ninteract with AI as soon as possible\num understanding how to speak with it\nhow to prompt engineer\nit's very useful and also it is actually\na fairly good educational tool right if\nyou could prompt it with right it's how\nI learn things it's extremely\neducational\num when I have extremely technical\nthings explained to me when I need to\nnot only explain to me but done for me\nit's extremely useful it is like better\nthan the young ladies Illustrated primer\nfrom you know Neil Diamond right no Neil\nSteveston\num by Longshot and so I've encouraged my\nown eight-year-old daughter to\nget chapter BT ask your questions teach\nyou how to use it teach you how to use\nit and she's like this is interesting\nthis is helpful\nso yeah it is\nit is you know people we talk about stem\nand getting young people to learn how to\ncode this is uh better than that it can\nwrite the code for you wow yeah that's a\ngreat point and it must give you peace\nof mind too knowing that openai is so\nstrict on not strict but so cautious\nabout safety as well yeah\nbeing able to hand over an app with\nonline it's basically on you know used\nto be worried about giving your kids\naccess to Google because they could go\nsearch for whatever but that's not the\ncase here but you know you do have to\ntell tell your kids that hey not\neverything is going to say is going to\nbe correct right and you need to like do\nyour own research and like look it up\nbecause it's going to make things up but\nit's good that's a good and valuable\ncurrency if you have with your kids\ndon't believe everything a computer is\nabout to tell you right that's going to\nbe important I think in the future don't\nbelieve anyone\non on the face of it right you need to\nlike think for yourself but this is also\nextremely valuable and increasingly\naccurate tool and what are some of the\nuse cases that your eight-year-old likes\nto use chat DVD for\num so she uses she programs in a Thing\nCalled scratch\num so that's been called Full uh she\nuses it to understand better she's like\ntaking like some uh at least for a third\ngrader so like more advanced math like\ngetting answers to the basic math is\npretty helpful\num I teach you how to like how she can\nuse charity BT to teach her things right\nso explaining my analogy explaining like\nI am in a year old because she is an\neight-year-old so we're stuck on\nsomething be able to formulate the right\nquestions and it's all all fairly\nhelpful to her yeah that's cool has\nthere been a response from the school\nabout what place chat GPT has in\neducation\nthat I haven't heard\num yet I mean she's in the third grade\nso there's not a whole lot to worry\nabout it but I mean in general if a\nschool is like don't use chapter VT that\nis like the wrong response right it's\nthe equivalent of like no don't use\ncalculators you should use calculators\nyou should also know how to add and this\nis something where it's not just a\ncalculator it can teach you how to add\nor divide and how it works\num and so\nyou know what I think a much more\neffective way of educating this is going\nto exist people are going to use it and\nyou should teach people how to get the\nmost out of this they're going to just\nrelying on it but making their own\nfaculties and skills and knowledge a lot\nmore better sure do you think that the\ncurrent education system the way it's\nset up is ready for this new type of\ncalculator for students to have\num I don't know I'm not super well\neducated on that thick\num generally there's been resistance but\nyou know I we live in Silicon Valley and\nI think there's a a lot more openness to\nnew technologies like this\nyou know like you said there are second\norder effects and we're barely seeing\nthe impact of that and I think the\neffects are going to be fairly dramatic\nfairly quickly right now it's a fun chat\ntool and you can talk to it what happens\nyou know what one of my one of my main\nquestions and opportunities around AI is\nthat\nprobably the biggest bottleneck and its\nimpact is the fact that for most use\ncases especially if you're using chat to\nBT\nit requires you to initiate the\nconversation\nand so a lot of the bottleneck is just\nremembering that you can initiate a\nconversation as opposed to going to\nGoogle or just trying to figure out\nyourself just initiating a conversation\nand so that's how a student would use it\nright oh I have a question I gotta\nremember how the talking BT so\nyou know it's going to be very impactful\ninteresting across a wide domain of use\ncases when that flips what happens when\nchat EBT or its equivalent is able to\ntalk to you proactively and explain\nthings to you right when you need it I\nthink that's going to be highly\ndisruptive to many many domains\nespecially in education\nand I don't think we're ready for yeah\nuh it's making me think of almost like\ncopilot functionality we're in co-pilot\nyou don't need to really prompt it to\ngive you something it's just going to\nkind of show you a shadow of what it\ncould do all over the place so I went\nover the co-pilot metaphor analogy looks\nlike around the rest part or the other\nparts of your life yeah right co-pilot\nfor your entire life right you know\nyou're reading an email\nfrom Greg and then it notices that you\ngo to your calendar and it notices the\nwords in there the kind of autofill\nmaybe it autofills for you who knows\nsomething like that yeah yeah there's um\nmultiple companies working on that I\nthink rewind is well positioned to be a\nplatform for that\nbut yeah that's where things are going\nright like the startup that I wish\nsomeone would do and I know there's only\n20 Stars working on this is you know\nRunway is at a place where I no longer\ncan keep up with everything that's\nhappening in the company\nand uh we also record every meeting and\nwe have email traffic and slack traffic\nand Property Management traffic and\nCommons and figma and\nAI could like take a look and be a\nmission about everything that's\nhappening in company tell me what is the\nmost important\nright that's that's just an Enterprise\nuse case but you can imagine it for\nevery other aspect of your life or just\nscratching a surface of all of this and\nI think per activity and changing\ninteraction model is going to be a\npretty large change I found that from a\npractitioner's standpoint as I've come\nacross problems like that understanding\nwhat's important and telling and trying\nto instruct the definition of importance\nhas actually been more difficult than I\nwould imagine too because that's almost\nvery human to understand instructing\nmachine it's been difficult yeah I you\nknow my own hypothesis on that is like\nyeah there is some element of\nunderstanding capability that's you know\ngpd4 is it still quite there yet\nwhat it has context about is about\neverything else in the world on a broad\nbasis but what's missing is content\ncontext specific to you\nand that's going to require a\ncombination you just need to integrate\nit with a lot of data\num and uh you need to give it more data\nthat's what's available right like I\nthink maybe 60 of the things it needs to\nknow about you is somewhere on in the\ncloud but 40 of it is something you need\nto tell it and um\nthat's still going to be challenging but\nit's a humongous opportunity again\nbarely scratching a trip itself yeah one\nof the paradigms I've seen other\nproducts do is as you interact with it\nit'll have a scratch Pad about your\npreferences so this is Greg's phone\nnumber this is Greg's email this is\nGreg's favorite ice cream or whatever so\nit'll be interesting to see what an AI\ncan pick up about us passively and how\nmuch watching our Behavior it needs to\ndo before it understands us pretty well\nright it's almost like the um remember\nthat Rochambeau automatic game or I\ndon't know it was the even odd automatic\ngame and you and you would input\nuh you would click a couple random\nnumbers or a couple random selections\nand then something after like 10 turns\nthe AI would predict with like 90\naccuracy the next move that you were\ngonna do so we may be a bit more\npredictable than we actually think\ntotally\nyeah totally\num so I want to bring up one of my\nfavorite tweets that you had talked\nabout recently and I told my wife this\nand she was almost didn't even entertain\nme\nliving forever\nso your Tweet had said\nif things continue the same way that we\nthink it's going to go anybody\nunderneath that retirement age today\nwill have the choice to be able to live\nforever or not maybe\ninteresting out of all of like the sort\nof predictions people have yeah of that\none I'm probably the most certain of\nuh that feels like the least\ncontroversial one that feels like just\nmathematical almost in my head\num and so you know you just have to\nbelieve a few things one is like\nyou know longevity is a physical world\nproblem it is solvable\nit does not violate in a meaningful way\nto laws of physics\nso you take that as one\ngiven\num and then you need to think about okay\nat what point do you have ADI ASI right\nnow consensus estimate on metacalculus\nis like uh 30 20 35 right and by the way\nthat estimate has been following Falling\nby like decades years every single month\nevery single year for the past 20 years\nso if anything it might it's probably\ngoing to come sooner than that but let's\nsay it's let's say it's 20 50.\nokay\nlike okay given where we are today in 25\nyears do we think we're going to have ai\nthat is smart enough to improve itself\nI yeah that seems fairly certain to me\nSo In and Out World what is unknown is\nare we able to control it you know\ndoomers say we can't or we're gonna have\nto work really hard to do it and the\nacceleration says yeah it's going to\nhappen anyway it's going to be great\nso either\ndoomers are right and we're all dead\nor the acceleration that's all right and\ndeath is solved and we're all going to\nlive forever yeah yeah I mean it's it's\nhard to argue with that with that logic\nso say you're presented with the choice\nhow what do you which pill are you\ntaking\num watch the good place uh no I haven't\nokay\num not a swollen anything but yeah like\nbasically people live forever\num and at some point after living\nforever for like a trillion years you're\nbored and you know you go to the next\nplace and you kind of off yourself and\nfind out what happens I'm in that camp\nyou know like I remember\num Aubry the gray in a TED Talk people\nwere talking about\num solving death and how\nsome people have an ethical issue with\nlike solving death\nand his responsibility this is like a\nhuge ethical issue\nand like why why is solving death a good\nthing because his answer was because\ndeath kills people\nuh I was like that's a pretty good\nargument\nright like no one thinks murder is good\nthat's the thing that kills people\nit's like yeah so it's bad and it should\nbe solved and it's like that's a pretty\nsimple ethical argument for why it is a\ngood thing to work on and people should\nhave the right to die too yeah you want\nto talk about second order effects I\nmean that is going to be\nmassive yeah that's why it goes back to\nmy point like short term maybe overtype\nlong term that\nit's like how do you and how do you over\nhype that\nyeah and I'm just thinking the entire\nworld right now was or has been living\nwith the expectation that they're going\nto die eventually there'll be a class of\npeople who will wake up be born and not\nhave that expectation and the way that\nthey think about life and the goals and\nobjectives and how they choose to live\nit\num I can't even begin to think where\nthat's going to go what other\ncontroversial thoughts do you have That\nOthers May disagree with you on\num a lot I'm not willing to say publicly\nuh specific to AI\num\nyeah I I think we talked about most of\nthem you know the biggest one was like\nthe idea of analogies\num\nI I have one so this is one we haven't\ntalked about it it speaks to the impact\nof AI and so there is a school of\nthought that says like AI is going to be\ngreat for creativity\num because it's going to be this\namazingly useful fun friendly tool for\nThought control for expression right if\nyou're a writer and it will make you a\nbetter writer if you're engineer or\nengineer artist better artist\nI think and so in that world human\ncreativity human input becomes more\nvaluable\nand I think\nmy issue with that is that that is true\nin the midterm but over the long run\nwhat it ignores is\ngoing to be funnier than you it's going\nto be more creative than you\nonce it's out in the world and it can\nthink and I think we dramatically\nunderestimate what happens then\nas funny as a creative director you will\nbe when you use AI to write jokes or\ncreate art what happens if AI is better\nand I think people don't really like to\ntalk about it\nbut I think that is to me fairly\ninevitable that it's going to have more\nfeelings in you\nthat it's going to know more than you\num and I don't think we're really\nmentally prepared for a world like that\nit's all fun and games they're going to\nbe a great tool and it's going to be\nhelpful for now but what happens after\nit's very parochial to think that humans\nwill always be looking down on the rest\nof the technology and the other other\nspecies\num we've never had to look up\num I was reading a deck which was a\npresentation by somebody I I really\nshould attribute I'll put in the show\nnotes but they had said that scientists\nhad created AGI on a Thursday but they\nmade a laughing like a comedy AGI\nit wasn't very funny on Thursday it\nimproved itself overnight it was pretty\ngood on Friday but it still wasn't that\ngreat so it left it over the weekend and\non Monday\nit was ten thousand times funnier than\nany human who had ever lived and the\nscientist upon immediately hearing the\njoke they died\nof laughter because they had it was\nincomprehensibly funny that their mind\nbroke and they just died and that was\nactually the destruction of Earth\nbecause people heard these jokes and\nthey just immediately killed them I\ngotta read that that's so good I'll send\nit to you over by\num it was quite long but there was a lot\nof really interesting predictions on\nthere as well another analogy I loved\nwas this author said you know we don't\nknow how um how intelligence is going to\nbe and the the the example he gave was\num Stephen Hawking's cat he goes Stephen\nHawking is an incredible scientist\nincredibly smart can you know talk about\nblack holes and all that but if Stephen\nHawking wants to get his cat inside of a\nbox he he can't do it he wouldn't even\nhe cannot coerce his cat to get inside\nthe box he goes that could be what super\nintelligence is going to be like or\nit could be like Einstein intelligence\nwhere Einstein is incredibly smart but\nif he wants to get his cat inside the\nbox and the cat isn't going Einstein's\ngonna pick up the cat and stuff in the\nbox and he goes we don't know what type\nof intelligence AGI could be it could be\ncould be one of the two I think that's\nright yeah who you know who knows about\nI'll send it to you after uh send you\nthat link to that one obviously anything\nelse or any other hot takes or anything\nelse you want to cover here while we got\nthe the mic open for it\num well in the short term it is going to\nbe the most fun five years that we've\never experienced in technology\num and so\nyou know I'm of account of like build\nand enjoy it and have a lot of fun\nwhile we're doing it so if you're an\nentrepreneur or developer an indie\nhacker or somebody who's not currently\nparticipating in the space what's the\nway that you'd recommend getting started\nget started\nyou know there's no\nreplacement for just going asks chatgpt\nhow to get started it's gonna have a\nreally good answers for you\num I wrote like a notion bot this is\nfire by uh John Cutler\nthis idea of how do you get Chaturbate\nto talk to you and of an ocean bot that\nnow will comment on any notion document\nin a voice of anyone you want it's\ncalled jam session at jam session with\nSteve Jobs and they'll just comment on\nyour notion doc says like extra people\nexcerpt uh comments and so here's what\ncpos would have done\num and you can get started it's never\nbeen easier to get started AI will help\nyou get started there's a lot that you\ncan do so\nuh just go just go awesome CQ well thank\nyou very much for all that you share on\nTwitter as well it's really fun to see\nyour thoughts your projects your\nphilosophies your predictions and I'm\nlooking forward to seeing more thank you\nI'll do more of it sounds good we'll see\nyou later see ya",
  "transcript_chars": 31570,
  "ingested_at": "2026-05-15T04:41:47.404766+00:00",
  "source": "channel",
  "yt_meta": {
    "view_count": 1735,
    "like_count": 51,
    "channel_id": "UCyR2Ct3pDOeZSRyZH5hPO-Q",
    "categories": [
      "People & Blogs"
    ],
    "tags": []
  }
}