Episode #414: AI vs. Human Potential: Stewart Alsop III and Alexander Explore What’s Next

Welcome to the Crazy Wisdom Podcast. This podcast is for you if you have an insane drive to find the truth of things. It's not the good answers that we seek, but the good questions. I interview a range of different guests from many different fields all with the intention to uncover the simple truths that are hidden in plain sight. Most people don't want to go there. I go there. My guests go there and you benefit. Please let me know if you enjoy these episodes and as always subscribe on iTunes, Spotify or wherever you listen to the podcast. My guest today is Alexander and we have been going back on fourth on Twitter. Well, I guess he's been going on Twitter. He's been publishing these great videos of him reading white papers and just like getting through the entire white paper and it sounded really interesting to kind of go in and investigate what what Alexander learned. So welcome to the show. Thank you. It's an honor to be here. Thanks for having me. So you talk a lot about AI and you talk a lot about kind of reading these white papers. What's the biggest thing you've learned in the past few weeks? Wow, in the past few weeks. So yeah, I've just been trying to get a general fundamental understanding of crypto web 3 because now I'm doing some smart contract engineering in the field. So yeah, I just really want to get my bearings. And so on that journey, I listened to a bunch of podcasts, because I love podcasts and like a total podcast addict and after listening to crypto podcasts for long enough, you hear people talk about the white papers. And they talk about them as if they're these magic document. The Bible. Yeah, it's like the religious text, especially big corners, because they those guys are just, you know, religion. It is literally a religion. And so I'm like, okay, let me let me take a look at these. Also, the company I'm now working for, you know, the engineers there recommended that I read the company's white papers, which of course, you know, if you're joining company, you should definitely read their white papers. So, you know, I want to read these papers, but I'm a Gen Z and, you know, years of terminally online video games, social media. You know, it has a real impact on attention span. And reading a white paper is like one of the, it's like a boss battle attention span test, in my opinion. Like I've done a lot of meditation. I do Zen meditation, where you just sit there for like an hour, like that's like the longest step meditative for like one hour, you just do nothing, focus on your breath. That is easier than reading a white paper. Like I would read the white papers and think to myself, like, man, maybe I should just meditate for an hour, because it would be more like doing literally nothing would be more interesting than reading these these white papers. I'm just completely sure that there's actually quite a lot of interesting stuff in them. But basically, I would sit down, you know, start reading the paper and then go, okay, let me, let me just grab some water. Let me go to the bathroom. Let me get a little snack. I'm a little out of focus. You know, maybe I'll just take a quick X break and, and then come back to this, you know, maybe I'll Google around a bit to get some more knowledge. And, and then I just like could not get through one. So, you know, the way I think or the way I like to live my life is trying to hack stuff. You know, I'm like, like, I want to take what's out there. gain a bunch of knowledge and find some unfair advantage like some work around, you know, and, and, and hack something. So, I was like, okay, I love X, you know, I'm posting on it all the time. It's addictive. Can I hack it in order to learn about crypto and read white papers and I was like, okay, well, how about this. I've seen posts before go, you know, someone viral go get popular. I've just people filming a time lapse of them working. Like, there's this one guy who would do all nighters and he would film a time lapse of him doing an all nighter and I would watch it every time because I'd be like, did this motherfucker actually go from like sunset to sunrise and you could see like he has a window. He's like in a basement suite and there's a window and you see the sun goes down and then the sun comes up and I would stay till the end every time because, you know, I just wanted to see if he did it. And, and staying till the end is what the algorithm really checks. That's that's how it knows to promote a video. Right. So that, yeah, I don't know if you want to talk about, I don't want to take too much of a tangent here. Well, I mean, I got plenty we can go from there. The biggest thing that you said was the, being a Gen Z and dopamine, the video games, the Twitter, online. It's funny because I think you actually have an advantage. It's probably harder to see from where you are now, but and I don't think it's shared among all Gen Z, but for the small subset of Gen Z, it's kind of like an advantage because you guys were so deep inside of the internet being online that it's just not even a question of how crazy addictive it is. But for everybody older, like the millennials, which includes me or the boomers or the Gen X, the boomers don't realize it. They don't have as much self-awareness. The millennials have more self-awareness because we saw it coming. But still, it's not to the same degree. We didn't, we didn't literally grow up with this sort of cyborg thing. I got the, I got the beginnings of it when I was six or seven year old, just because I grew up in the Bay Area and it was around technology when it was first starting, but totally different. And so this is really interesting. But also, I've tried to use Twitter in a similar way back when it was less effectively run, which was actually great because it allowed me to do these stable threads. I don't know if we were connected when I started doing these stable threads, but I think I started back in like 2016, 2017, and I started using Twitter as my personal note-taking device. So I noticed that it's a great note-taking device and it's social. So sometimes your notebook like talks back to you, which I found it's so cool. It was, it was great. And I did these stable threads and it was like on one topic. Every time I got a new thought on that one topic, I would write about that, that new topic, and it would just be this thread. And then Elon Musk comes in and just destroys the algorithm for this thing. So they're not being seen anymore. My notebook doesn't, doesn't really talk back to me anymore. Now I think I'm completely destroying my engagement. I'm not sure what it is exactly I'm doing, but my engagement is just completely flattened out. And so yeah, it's really interesting. So I love the video idea. I love that sort of like having this, because we are such social beings, this myth of the individual, although I think individualism is very important. There's also this myth of it. And like we're so mediated by our social environment that it's really a great hack to basically hack into that expectation accountability thing. So I think it's a really interesting. If you have anything you want to respond to in that feel free. Otherwise, I have a bunch of follow-up questions about what you learned. Oh, yeah, for sure. So yeah, then I, you know, because I didn't fully get to that end of the story. So then I decide, okay, I'm going to film a timelapse of me reading a paper from start to finish. And the challenge will be to not leave my seat and not open Twitter, not go on my phone. My phone is filming a time lap. So I can't use it. And then yeah, I posted it. It got like 130k views, which blew my mind. Like the most views on a tweet I ever got was maybe like a couple hundred. And the response was so positive. Like people were so encouraging. Actually, there's a couple of haters, which is really hilarious. Like people were, like a couple of people commented, like, oh, is it an accomplishment? Like do you think it's an accomplishment to read? Is that how bad our society is? And I was like, yeah, bro, look around. It is. Yeah, like, so that was that was pretty funny. But then it just created a, like a virtuous cycle where I was like, okay, let's, let's read another. So I read two more in the next two days because getting 130k views is such a massive dopamine head, you know, like I'll totally chase that, right? Like all like, you know, a thousand likes. Wow. Like that's, you know, like the, like your phone is just buzzing constantly with all these likes and comments or whatever. And so yeah, it made, it gave me a tremendous amount of motivation to read white papers. And then I went on read like five more. And I just think that's really fucking cool that like you can go. Not not exactly viral, right? Like I consider viral like maybe a couple of million views. But like, you know, like a hundred thousand people will watch you read a paper. And, you know, like 60 of them will take the time out of their data comment some type of support. Like, yeah, you got this. Keep going. Like that's actually really cool. And X is a mountain. You're really underrated app. Yes. Very interesting. There's more we can go into on that the sort of distributed cognition and not only distributed cognition between humans, which has always been the case. But, you know, you bring in writing, the invention of writing about 2000 years ago, 2005, 500 years ago, maybe longer than that. And then all of a sudden you have distribution of cognition socially, but also among time. So that it's no longer f a mirror. It's stable throughout time. And then you bring it today for not only do we have writing, but we also have artificial intelligence. And we also have the internet. And we also have blockchains, all of which completely changed the nature of the distributed cognition. But the most potent is still the social media, although it's going to change a lot, we could talk about how you think it may change. But there's there's, but that also want to learn the like what what is the most technically surprising thing that you learned from reading these white papers? Like there's the the sort of dopamine and everything, which is really interesting. But is there anything that you just really caught your surprise about the way things work in any of the blockchains or in any of and I would love to understand more about the chain link one if if that spoke to you at all. Yes, so I guess the what's interesting is how simple Bitcoin is that was something that really stood out to me. It's just a steers of very simple ideas that was very hard to put together at that time and figure out at the time. Yeah. And it's just so well explained. You know, I got I get why people treat it as this holy holy script because there is a beauty to it. It's like elegant. Yeah, it's very elegant, especially after reading a couple of them and more modern ones. Yeah, it just really stands out as as like a work of art that it gives like Oppenheimer feelings. You know, like you gotta be playing the Oppenheimer music in the background while you're while you're reading that paper because it feels like you're you know, you're precipice of something and you really are. They created a trillion dollar asset out of out of computers, right? Like that that that's wild to me. Yeah, and so yeah, so the way information is stored in Bitcoin was really impressive to me. It's so simple. You would think that for Bitcoin, you know, to own a Bitcoin, okay, maybe you got to set up an account. You have to, you know, send like a message to someone else. You got to have a profile that you have to have a central database that keeps track of all the money and how much everyone owes each other. And what it really is is like, you know, you hear about the ledger, the blockchain. It's just a list of transactions. There's a lot going on in terms of like mining, validating data itself, but the data itself itself itself is so bare bones. It's just transactions to addresses. It's it doesn't even say, you know, address, like Stuart's address has 100 Bitcoin or something like that. It just says, you know, address a cent Stuart's address 50 Bitcoin address be cents to its address 50 Bitcoin. And you know, it just knows from like it. And it just has nothing else. It's it's so minimalistic. And I really think the smartest things, the most beautiful things are the simplest ones. You know, the simplest ones that are easy to explain, easy to understand. And there's a genius in simplicity. So let's talk about AI from here. I heard a long time ago that smart contracts were actually the most applicable to AI because it's all in a blockchain. So it's all readable. So it's not like there's a bunch of closed source applications that are outside of the training data that the training data for albums is is for the most part has a lot of smart contracts in them. Is that accurate? Yeah. So all smart contracts are publicly readable. They're deployed on the blockchain. They are deployed as bytecode. So it's, you know, there's if you download, let's say all of the information stored on the Ethereum blockchain, you're just going to have a bunch of bytecode. And bytecode is like code that only a machine can read humans can't. I mean, I got it. Some crazy people out there might be able to read it. However, because it's out there public, the repositories are always are very predominantly open source and readable by anyone. And they also do audits to ensure that and then also contract verification. It's called. So basically when when you have a contract that's written in, let's say solidity, solidity is a very common programming language for smart contracts and it's human readable. It looks kind of like another language called TypeScript, which is very commonly used in web development. So if you write a contract in solidity and then deploy it to the blockchain, it's going to be compiled as bytecode. Now compiled means you take this human readable code and a machine turns it into machine readable code. Why do we turn it into machine readable code? Because there's a lot of optimization that goes on in human readable code. You have a lot of extra fluff, right? A machine doesn't need to, you know, like, you know, what the name of a variable is or something like that. The human readable code, you got to come up with a variable name that makes sense for human to read. You can't just name everything A, B and C, really minimalistic thing. So it gets stored as bytecode, right? And then someone can interact with it on the blockchain. So let's say I want to call Stuart's contract that, you know, like, invests my Bitcoin and gives me like a yield to it. Well, I can't read bytecode. So let's say Stuart tells me, hey, I made this contract. You can give it some money. It'll give you a yield. Then you can take out your money. I just have to trust you because I go, well, it's bytecode. I trust Stuart. I think he's a great guy. He's got a good podcast. He's not going to scam me. However, that's not what like the entire point of blockchain is that no one has to trust anyone. So there's a thing called contract verification. And what you do is submit your solidity written contract that's human readable, a service that's managed by the blockchain. They do the compilation and then compare it, compare that bytecode to the bytecode on chain. And then if it matches, the contract is verified. And you'll have a human readable representation that's verified. Then the bytecode living on chain that is, you know, basically this entity, this transit entity has said that they matched and confirm it. I got a question. I got a question. The bytecode. So you're comparing two things. Well, you're comparing the smart contract that's converted into bytecode onto the blockchain. And you're comparing that bytecode with what? With bytecode that is compiled basically in the same manner, but it's not on chain. So how does that, so okay, so this is great for me because I'm learning a lot here. So there's a smart contract that's between two parties, usually in solidity. And we don't know each other. So it's sort of like trustless. We want it to be trustless. Then the smart contract is then written in solidity, which is human readable. And this gave me an interesting insight into code in general, because it's basically not just for the machines, but it's for other humans also working on that code to both know what's going on. But then there's this other thing, which is that it's trying to interact with a blockchain. And so how is the bytecode verified by the blockchain? Like this may get into like the actual virtual machine on Ethereum if we want to go into that, but the like what's going on in those places for lay people. So I'm not actually sure the specific details of verification other than some party managed by the blockchain will do that verification process. And so basically what you have is so Stuart makes the contract in solidity, uploads it to the blockchain and it's compiled in bytecode. No one can read that by code. So we have to trust Stuart in order to interact with his contract. However, Stuart can send his human readable contract to some group managed by Ethereum. That group will compile it into bytecode compared to what Stuart says he's put on chain. If it matches, then it's verified and that's stored somewhere so that other people can can see it and know what code is being run when they execute the smart contract that's unreadable to that. It's now read and certified or verified. That was really helpful. And so now taking it back to AI, how does that? How is that? And we want to talk about AI in general as well and how software development has a skill might be sort of not necessarily the best thing to invest your time in right now. How do those two things relate? So for these large language models, you need to train them on data. And it's really just about the quality of the data, the amount of data you have. And data sources are extremely, extremely important. So if you want your LLM to be really good at coding a specific type of software, it's got to be able to see a lot of that software. And because all the software that goes on the blockchain is public code basically and then it gets verified and it's open source that's put out there on the internet so anyone can read it. Just everyone wants their contract to be used and interacted with. But no one trusts those contracts so they have to get verified, which means they have to post their code out there on the internet. That means that these LLMs have a ton of smart contract training data out there and basically can train on all smart contracts ever written. So yeah, LLMs should be wizards at creating these smart contracts. And I've definitely felt that they are a very helpful tool in understanding smart contracts, architecting them. Yeah, just coding in general. And I would love to talk about where it's going. Yeah, where is software going? So I actually I want to turn it around on you if you would. Yeah. And then I'll wait in. So basically, I mean, I've now been running the AI. So I got a job at a company that was one of the main vendors for OpenAI. It was only a few months after and quite strangely, serendipitously ended up at this company right after CHAP GPT was announced basically inside this company. It was growing massively, going through hypergrowth, all being a vendor towards to OpenAI. And so and I got into this company and this is a really interesting company because all the people at this company were experts in machine learning models. Lots and lots of really smart programmers who were implementing the details of machine learning models. And there's a whole bunch of business process outsource people. They were doing a particular algorithm that required a huge amount of human labor to go over JSON and go over a bunch of different inputs to make sure that all of them were accurate. And so had this really interesting experience where I'm surrounded by all these people who know all these juicy details about these machine learning algorithms. But I was teaching many of them how to actually use the LLMs to become productive. I was leading workshops with it because something about language with the LLMs allowed me to really get it really quickly intuitively inside the chatbot. I'm not talking about the API. I'm talking about the chatbot experience. And something about that language and linguistic ability allowed me to like maneuver the language of it to get stuff done inside of an organization. So I started teaching these workshops. I ended up moving on from that company and I started to here in Buenos Aires started a weekly meetup called the AI Whisperers where I tried to find other people who had the same gift as me to basically linguistically get what I was looking for. And over the past two months we started to do workshops where we're teaching totally non-technical people how to get the AI to prompt the AI to give it code and then get that code to get run. So I'm not we're not teaching anybody how to code. We're teaching people how to get the AI to code for them because my thesis is that and it's been proven as soon as 3.5 came out I was like oh this is what I need to finally start coding and then I realized okay this is what I need to get to do the coding for me because I'm never going to be better than this thing at coding. It's always going to surpass me at coding. But if I know how to maneuver it and get all these things together then it allows me to take advantage of this thing that allows me to leapfrog the necessity to code. And so then I started in so in the last few months we started I've been sharing what I've been learning about coding and that's just keeps on I keep on being proven accurate that this thing is getting better and better. Went from chat GPT 3.5 to chat GPT4 to Claude Sonnet 3.5 and now two days ago repilit releases their agent and all of the things that I've been teaching these people these non-technical people how to do to get their code deployed to the website. I just throw throw one prompt into the repilit agent and it does that exact same thing that I've been teaching these things these people how to do at this repilit agent just goes and does it and I was able to deploy it and it was actually it's we got them we got all the humans to deploy this a week ago but it was a significant challenge to get it to do a API call from a hosted deployed website that was like a really big challenge that only about 60 percent of the the people in person got it and then with five minutes this repilit agent does that and to place it and and so I haven't even checked that out that's crazy it's I I went I went from an agent agent bear to an agent bull I now think agents are here and I think agents are really important and I'm also starting to realize that we might be significantly close to autonomous AGI like because if if this thing can can can can build applications and this is just the alpha version of it like in six months this thing is going to be building applications and and I think there will be a lot of edge software that really does still require a lot of human beings to solve something new you know an example would be I can't I don't know what's going to be in the future but an example would be Facebook in 2006 or 2007 trying to do a entire machine that maintains state for a billion people like that type of technical challenge is still probably going to be humans for for the next two years at least but anything outside of that is going to be able to do that it feels like the day I is going to be doing what do you think do you agree do you disagree anything you want to add I completely agree with everything you said I'm so glad I turned that around on you because I did not have said it better than myself and also while you were saying that I was thinking a very interesting question and let me think on how to raise this something like so basically the what I'm thinking is we're almost in competition with AI as a you know with all of these chatbots yeah basically yeah so like as a human if you want to do a job yeah you're going to have to compete with the chatbot so let's say you just started learning something like for you it's coding can you learn faster that you can be better than the AI basically so if you started learning computer science today and you're in your first year of university god help you if that's the case particularly with what they're teaching because they're the style of teaching inside a university is the exact opposite I would I would if I were dictator and immediately had control over all of the universities in the world I would teach the exact opposite every university is currently teaching like I would say use this AI as much as you can learn how to navigate this thing as much as you can so I if you if you hold on to that thought because I want to give a little bit of a nuance into what I think which is essentially that it's not necessarily competing if you're talking about the individual skill like coding yes 100% I am in competition if I tried to learn how to code but I would also fail that competition so I guess I would like it's just like I've already given up in that competition because I just know that I'm I know the limits of my competency and I'm not going to learn how to code better in Python in six months by the time it's better way way better at coding Python but the human being is beyond the linear understanding of the of our wealth this gets in a little metaphysical or conscious aspect the linguistic and spatial feeling I have of me Stuart sitting in this room is a little bit of a of an illusion because under my understanding reality is combinatorial explosive complex to the infinite degree and the my little understanding is this really really helpful evolutionary piece that allows me to make all of that complexity sort of manageable and but that's not who I am this this conscious linguistic is an important aspect of who I am but it's not who I am and that who we are is is I want to say removed but it's just not it's in the same way that reality itself is complex who I am is also complex to that same degree because I'm part of reality and so we only are aware we human beings evolutionarily are very much a problem solvers since we're constantly trying to solve problems out there in the universe and we're sort of limited by the problems that we have in front of us like my imagination I can't you know I try to think about what's going to happen in five years with all these different things but my imagination has limits in in place which is that the current environment is constrained when I have to deal with those constraints and and my imagination is somewhat blocked by those a lot of those constraints are about to be removed and so what does the human do well the human has this beautiful imagined imagination which has sort of been kind of disregarded for the past 60 years as we become hyper focused that imagination I don't think the AI can access that I think it will access parts of it and I think it will really challenge us as to what we think a human really is but I don't think it's going to get to the whole thing so I still think that humans are going to play a very very important role in this but it's not going to be the role we placed we've played for the past 200 500 thousand years but two thousand years basically wow yeah it does the conversation just converges on what is consciousness at the end of the day you know that's that's what it comes to and it's just a reflection of how insane the times are that we live in now like I always talk to rumors and I say like I think we're living in the craziest time to be alive and they'll always tell me oh that's what everyone thought you know that is different in the evil times the the wagon came out and everyone thought the world was no man it is different yeah it's called the and I would like to know I'm not I'm not sure if your listeners are familiar with the idea of singularity like technological singularity wait if I describe that would that be a waste of time yeah absolutely everybody's heard a million times okay yeah then we don't we don't have to talk about that but well no we can't we can't talk about because I would love to hear what you're going to say about like as well singularity is really interesting I'm not sure if you've seen my tweets recently I'm somewhat critical of the the singularity thing because I think it's part of the third grade awakening in the United States a religious movement I'm not sure if you are familiar with the first or the second grade awakening inside the United States they're sort of arcane religious revivalists movements the second one led to the Mormons and I think that the 1960s where the singularity kind of stuff got got started is part of the third grade awakening inside the United States which is I the singularity stuff feels like a new religion and it's a millennia naryanism religion in the sense that there's an end times but in the singularity is interesting because it's almost like a utopia end times but it is an ending for humans if you have anything to say but I'm very open to hearing other perspectives of the singularity I don't mean to discount it or anything like that what's your take there well so yeah that's interesting that you think it will be a utopia that's cool like I like to think optimistically and there's really really no point in being a doomer other than having a healthy amount of cautiousness but you don't have to be a doomer to to hold that um but you know I think I just like the black hole analogy um a singularity of a black hole is where you're sucked into it and there's no return yeah and also the part that is talked about less frequently is that you cannot see inside of a black hole light does not stay so you cannot see past the bounds of the singularity and I think the way that applies to the technological metaphor is that we have no idea what the world will look like because it it is a singularity it but you get sucked into it like technology reaches a escape velocity that we cannot we cannot predict what will happen with levels of intelligence that we cannot comprehend and and so this does kind of bring the boomer comment back into it which is like isn't that already the case not in terms of technology like I agree with you that that this technology will defy any sort of ability to understand without being part of it without in that moment five ten years from now like we we won't know what that entails we can't know that entails and like if we're a omniscient person sitting on the outside it'll just look like a bizarre like the depictions that they give of angels where it's like just this bizarre thing your mind can't really put it it's just like oh that's weird um but uh the same it's like metaphysically we're always in this moment of judgment where it's I have no idea what's going to happen the next month I've got a relatively I mean in next minute I've got a relatively good idea of what's going to happen the next minute or two minutes but particularly in a podcast like it's extremely difficult for me to know where we're headed in five five minutes um like the basic you know the who I am all these like kind of like habitual things like all that seems but then death comes in like the death of the physical form like beyond that is a gate at which I know nothing and it's pure uncertainty and so that's already like a human element that we deal with and that we've been dealing with thousands of years is that there's a there's a gate which exists which destroys the perception and at that gate like we do not know there's all these theories about what happens so it's it feels like yes it is different but at the same time it is kind of the same thing we've we've been dealing with this entire time and that the tools that we've had for thousands of years are very applicable in the but but as we've said the question comes right back to consciousness which which has not been a question that seems very important for a lot of people for the last hundred years like 50 years like yeah yeah it does but I would argue that so death for you is uncertain but your death for me is not uncertain so if so the way you comprehend your own death you go okay there's there's death and then I don't know what comes next for me comprehending your death I go stored is going to die but I'll still keep on going. I know what will happen next his body will be there will bury it it will sit in the grass and decompose and then I will continue on with my life but I'll be missing Stewart and so but then for my own death I go well what's coming next well will I you know move on to another realm with my consciousness taken out their form will I join the super consciousness of the universe or will I be reincarnated will I you know burn in hell for eternity but so yeah there's like but for technological singularity everyone is experiencing the same kind of death rebirth cycle as long as they plug in right because it's like in that in that in that merging because I don't see how the singularity doesn't happen without merging but I guess people could argue that the merge has already happened we've been talking about it this entire time like we're hooked it on to this online world we things happen in this online world that both of you both you and I experience all the time but an Amish guy in the in the woods is like totally outside of it just like all the events through which we give significance don't really happen and so like we could there could be an argument that's already happened we're already merged but there seems to be a deeper level of merging that will happen as well and that for anybody outside that deeper deeper level of merging with the technology it will just appear as the same thing as the like it just becomes nonsensical to discuss what happens on Twitter with the Amish people in the same way it will become nonsensical to discuss what happens in this crazy AI enabled cyborg world basically yeah I just think we don't know if it's even possible to sit this one out you know are that you can just so drastic that affects everyone yeah yeah so and then yeah that's why it's is it there's a good question is it positive is it going to be possible to sit this one out so let's go like as AI we like we can both of us are enough inside the AI world to understand a little bit of the flavor of what's going to happen so in the last few days I've started to understand that it's the like just like that doom demo where somebody just trained in LLM I think it was Google trained in LLM on doom from the 1990s the video game for the 1990s and then instead of having a whole generated world that the world was generated on the fly and it looked exactly like doom so you take that you take generative graphics and you take generative stories and you take generative images and you take generative music and you combine all those things together and then you put them on the internet and also potentially down the line on VR and AR like that's going to create worlds but how do those worlds kind of develop it feels like I did an interview yesterday about they're going to generate based on customization and how does that change the internet because so far the internet developed under the context that were sort of in this one globe and and and a global perspective and that was very much from the 1960s with the the space flight and the and you know seeing the earth from the outside perspective and that created the internet or the internet created was created after that and then the internet fractured everything into a whole bunch of different worlds of interest and now you add that AI generated aspect and it feels like it's just going to create sort of new worlds that are customized based on our own feedback feedback systems do you agree with that do you disagree with what's your take there I would argue that a social media feed is a world yeah yeah and that is exactly what's happening right now you have an AI curating a world for you generating it basically the content is created well now it's starting to be created by AI right that we're you know one in every who knows how many posts is completely generated by AI and we're converging on yeah full world generation where everyone's in their own like oh man are you worried about echo chambers well you should be worried about you know your own chambers yeah because again it and this goes back to the the the tools that we've been given over the past thousand years as it's always kind of been that way but now it's just going to be with napalm on it it's like yeah and so how do we stay sank yeah that that's a good question I touch grass man I exactly yeah dance laugh have fun travel don't need to travel but stay in nature like breathe a bunch a lot there's a lot of tools out there again it goes back to thousands of years it's like we've had tools we're not like like the world as it existed for the past thousand years is also not been a utopia it's been quite intense and quite challenging for a long time so and I question whether will be a utopia I don't like in this infinite generated worlds it'll it feels like it's going to be both utopia and dystopia depending on what you're choosing sort of like a new age hyper reality tunnel type of thing yeah and question is do we just stay in our own world with our AI generated friends or AI girlfriend no or do we enjoy these worlds together yeah this is yeah this is a really good question for you specifically because you're part of Gen Z and so Gen Z is I think most liable to both breaking free from it for a small percentage but also for the larger percentage and not at all even questioning it and I'd love to get your perspective it sounds like you're questioning it a bit yeah I think I have from this conversation I imagine that you're you're probably going to be one of the guys who gets out of this in a in a sane way but what about your compatriots what do you what's your what's the sense you get from people in your age group I think that a lot of people in my age group are very behind on AI and unexpectedly so like you would think a lot of us would be super tuned into AI knowing how knowing how it works knowing the landscape what's what's coming out but it's actually like quite downstream people use chat GPT but that is about all yeah we'll do like know even really generating images like maybe some artists are but now AI usage is is quite lower than you might expect so I think it's just gonna hit my generation by storm I think people don't even understand that an AI hyper intelligent AI that's tracking everything they like and swipe on and comment on share you know pausing on a certain video I don't think they realize how what either being tracked and observed and catered to and because the AI is just so abstracted and a funny quote or a funny thing someone someone told me about tiktok this girl I was on a date with her and she said I started using tiktok and I didn't like it but then I figured out how to use it and now I love tiktok because I can use it to find you know my favorite restaurants or recipes and this but it was like you do not figure out how to use it you use you people are still thinking that they're in control of these things that they're oh you know I'm but it's completely manipulating them and and so I think yeah you know like you said there's going to be a small percentage of people who are who understand what's going on and can utilize these technologies and build on them and hack them but the majority of people are just going to be swept up in it and enjoy being swept up in it. This reminds me exactly of Jacques Allouil who is a French philosopher from the 1960s I believe Christian anarchist believed that the way technology called it La Technique and La Technique has this this the way the technicality of modern civilization which breaks everything into efficiency and we've got to get this efficiency going and and and it doesn't matter any of the other variables don't matter it just matters how efficient we're becoming and so this and this was back in the 1960s this is before software engineering before like all these different things and there's this really interesting sort of dominate versus submit type of thing and and that the technology appears to be submitting it appears to be a submissive force but it's not it's a dominating force in this weird thing and and and it also kind of relates to politics as well and that that where it's no longer centralized politics like 150 years ago you could have one centralized leader it was kind of decentralized because Stalin wasn't just Stalin he had all of his his group of people that he was depending on but then he eventually killed them all and but then there was a new group of people who came and supported him and then they were always scared and so like there wasn't complete centralization but it was far more centralized if we look at the political parties a lot like it's not really a centralized thing anymore it's a decentralized thing and you can see this in the 1960s as well with their French student revolutions which also were similar in the United States where it was like because of this TV because of this new model of of of reality and like you know the idea of a middle class and they all like it allowed for people to coordinate in a decentralized manner and so this lot technique this this this this this forest and civilization is is challenging because it we're totally so focused on these on the little tiny details and we're forgetting the giant forest we live in and it's going to be crazy and crazy with technology and technology in with the appearance of submitting is going to become more more dominant for most of the people unless and this is where sort of Nietzsche I'm not sure if you're familiar with Nietzsche but Nietzsche kind of his will to power is an important understanding to to those of us who don't want to just go gently into that dark night of of weird you know idiocracy and want to create something beautiful and and and and contribute to a flourishing of humanity because I do like I'm an optimist I think that on the long medium turn that this is all going to be very good for for human flourishing but could also be quite crazy in the near term and yeah what's your take there yeah it's just quite quite insane times and and sometimes I just wonder if it's even worth it like should we just be kind of hanging out like we're in a waiting room right now and it's not fun to be in a waiting room no and and so so another thing I wanted to touch on that we were talking about previously is the rate of learning so yeah and how we're learning so if you start learning a subject now because you want to acquire a skill that can go out and make money so in one year will you be better at that skill than the like chat GBT5 because in a year that will be out you know it might be out in a couple months actually and then in two years will you have more skill than chat GBT6 and so on so it's like can at what point does the rate of human learning get eclipsed by the rate of AI learning and what does that mean learning in general but should you even learn something if you'll never be good as good as an AI so this is actually directly related to my strategy for learning and the strategy is the same one that I used on Twitter I also used space repetition software in order to do things the one key thing I think we're missing in this conversation is that the prompting something requires an understanding that that thing exists and it also requires a meta understanding of the thing and I'll give you a direct example that's really related to what we're talking about with AI whispers and with teaching people how to code who don't know how to code including myself I don't know how to code and so when something goes wrong you if I had you right there you would be able to figure it out probably in about the attempt attempt of the time for me I have to go through all these different things that the AI sends me of like well it could be this could be that could be this could be that and then I have to research all of those things to get an understanding and this happens over days so eventually I do figure that out but but like there is this there is this so I don't in the in a short term I think that software engineers have a huge arbitrage opportunity like massive massive arbitrage opportunity I like we put that arbitrage yeah yeah yeah what do you think like because that's that's kind of contrary to what you just said about learning but only a short to medium term yeah well I think right now the good software developers have AI superpowers yeah exactly so I can code now in any language any language just throw it my way and I'll be able to read it code it debug it because if you have a certain fundamental understanding of software engineering and you've you know you're comfortable you've learned one language really well you understand you know compilers you understand under the hood what's going on yeah memory is used you understand a bit about object-oriented programming how to structure a code base you know kind of the basics of computer networking and cloud storage cloud hosting that's that's it that's all you need to now go ahead and use AI to build pretty much anything you want and and there is a phenomenon where AI creates code that it cannot debug yes yes this is I have to keep on running into this with the with the agent as well with both cloud and the agent that's when I know that I have to go learn something is when I can't get self basically yeah it's it's I think it must be some kind of love entropy or yeah it's thing like that that it's it's running up against but yeah it'll get into these loops where it'll tell you it knows how to solve the problem but it it's just stuck you know and it eventually it's not even changing the code at all and almost gas-legging you that's that the code is fine and that it's fixed it and kind of giving you the same set of you know 10 steps to take and they they just over and over again as if that's gonna solve the problem and that might be also due to its limited memory that it can't remember uh-huh uh-huh uh-huh what it did what it did which is yeah which is the the craziest thing because I think so many people on Twitter specifically are sort of fooling themselves about the consciousness of this thing and whether it can do reasoning I find that the dooms are very very interesting to kind of peer into their psychology at the same time as they're trying to take these little hints from the psychology of the LMS to say okay no it's gonna kill us all like it's too powerful and then like have you tried to get it to code a piece of software for you like what's like it doesn't it doesn't do it it's like it needs a huge amount of guidance still uh-but again it's all like you know who the future is is upper grabs in in terms of what's actually gonna happen so they might be right but um uh if you were a last like five to ten minutes or so if you were a junior developer trying to get a job right now well actually first let me ask you are you would you consider yourself is there are you would you consider yourself a senior or a junior or is there a medium engineer is there a mid one what what where do you put yourself on that perspective right now that's hard to say because with it it's just hard to define that and there's any different specializations in software so I'm a junior smart contracts engineer I've only been doing that for really a month in terms of creating like a SaaS cloud hosted product I wouldn't not call myself a junior engineer yeah um definitely not a not a senior engineer but yeah it's you know that whole like comparison between it's hard it's a good offer to developers yeah it's getting harder to compare because if you are using the latest tools you can fill in a lot of your knowledge gaps quite easily it's like being able to ask a senior developer everything they know about programming in a certain language and just having them sit with you all day like if you have a senior developer sitting with you all day that you can ask any question you want and you already have a good amount of fundamental understanding yeah yeah it's you know what what level of developer does that make you and the person who actually suffers the mo it is arbitrary because the person who not exactly suffers but the person who benefits the least is the senior software developer they're not benefiting from a AI at all right wait why I thought they would be the most live oh because they're sort of they're because they call themselves a senior software developer that they already know how to do it so they're the ones least likely to use AI why what do you mean there exactly they don't yeah it doesn't really help them yeah maybe if they wanted to transfer their skills to another area it could translate their knowledge of a certain framework in language to a different programming language and area however in their domain yeah they they write code they they debug easily they understand these very large systems and they don't need Chanty PG right they don't need a clot and and so but now for those senior developers at least the ones who use AI and it was so interesting for me to originally easily intuitively understand once 3.5 came out that this was the answer that I needed for myself to finally get across this bridge of coding which I did not like and then talk to so many engineers over the past two years and and be like just have a total non-curiosity that's changed over the last few months now a lot more when I asked them are they using AI are they using cursor are they using these these tools they're like okay yeah we're using this and so these these the senior engineers they don't need to hire junior engineers so the question I was going to ask you before is if you were a junior engineer if it is your first job getting paid to code you just got done with a computer science degree or you've been teaching yourself that's another good question is whether you think I know the answer between self-taught and and computer science who's going to benefit most from AI also depends on the individual but but if you were a junior developer well what would you do in order to get a job do you think it would be totally hopeless because that feels like like it's just like a senior developer is not going to hire a junior developer anymore because they're just going to be like oh well I just use AI to do this thing rather than add a new human into it which has all these other costs to it what's your take there yeah if so I think interest rates being higher has caused the tech job market to get kind of crushed and it's we're at you know pre-COVID levels of new job positions being opened and software so there's kind of that macroeconomic phenomena going on and then as far as AI goes the senior developers wouldn't replace a junior dev with with AI yet there's not enough trust there you know it still has to be kind of managed but there are much less junior developer jobs going around right now it is it is really tough out there I know several amazing programmers really solid engineers who cannot find a job and there's a there's a oversupply issue of software engineers right now so if I was a junior developer what I would do is try to be an entrepreneur and that was my that was my plant I looked at the market and I was like oh my goodness it's going to be a dog fight to get a job at you know in big tech or just add like a really good software company so instead of trying to apply for those jobs I'm going to use AI to attempt to create my own because we're in this amazing time right now where if you do know software and you have product ideas you can go ahead and build them faster than ever for cheaper than ever so that's what I'd recommend doing try and build your own product you're going to get a ton of super relevant experience on the way that can be applied to a job if you build in public and share what you're building you're basically advertising yourself as an engineer to these companies and you can attract recruiters or you know people looking to hire engineers so basically worst case scenario none of your startups pan out and you know you're doing Uber on the weekends to keep the lights on middle-case scenario you get hired at a tech company and you get a job because of all this new experience you've acquired and you've been building in public and you know you've kind of you've networked and found your way into a good position and then the ideal scenario is that you've created an amazing product that people like to use and you've leveraged AI and now you have a startup that is you know getting some revenue and get some investment and you can even work for yourself yeah that that leaves the sort of glaring hole in our conversation is distribution and human to human relations which is so far from all the technical details that we've been talking about so far and it's so much more subjective and it's so much more weird and it is learnable but it's not learnable in the same way as engineering it requires this social intelligence that is is really fascinating and it's not going to go away like it feels much much less likely that the AI even touches this business development human relationships those types of things do not feel like they're I mean I think there are a bunch of little tiny jobs within that that are going to improve but the actual just when you're relating to a human being in a commercial transactional or transformational way that involves money it's just such a large other plane field that AI is going to help with but I don't think it's going to touch in the same way that software development is going to change significantly but I totally marketing is might be one of the the last things AI does really well and sales too because it's not just marketing yes the the the actual like why does somebody buy something how can you influence them to buy something what are all the the what are all the negative externalities of buying and selling things where all the things that go wrong where all the things that go right like so subjective it's so weird it's so much based on our sort of monkey or among like our monkey evolution but also not just our monkey evolution but about beauty and competence and all these other things that it's just a totally different world which I I keep on going back and forth whether I want to do that or whether I want to be technical or whether I want to go explore deep tech hard tech go get on satellites or put satellites in the space or space factories or nuclear power or all those things which are also coming down online may not be as quickly as ADI but also might improve but this has been so much fun how can people find you if you're writing or if you do want to share you don't have to share but how how can people find you on the inner website uh yeah I'm on x as at Alexander tweets and instead of ease for the tweets it's threes so tw33ts at Alexander tweets thank you so much all right thank you Stuart thank you for listening and I hope you enjoyed this episode as always you can find me on twitter at Stuart all sub i i also don't forget to subscribe on spotify or iTunes for every weekly episode that I publish on Monday mornings hope you have a great day