speaker-0: Excellence needs no justification. This AI stuff is super, super easy to learn. If you're a very good programmer, it's a weekend. But you shouldn't like invest all your chips and getting good at that because there is no good at AI.

speaker-1: There is no good. There is no skill. And that is something that I I that that really gets me gone.

speaker-0: Like I'm not gonna hire someone 'cause they're good at AI.

speaker-1: People in my comments are telling me it's a skill issue. You're telling me I don't know how to prompt. You're you're cooked. Like you have no idea.

speaker-2: Welcome to the Based Academy. We are your hosts, BigBoxwee. Hello. And me, Speak Machine. Hello. Our guest today made a video recently, which clearly struck a chord with the programming community. His video called I'm Done Coding with AI has hit half a million views within two weeks, and it was already at 250,000 views within a few days. The implications of this video and the fact that people feel so strongly about it, I think are worth discussing. And so please welcome

speaker-0: I have to ask where is that last name from?

speaker-1: Yeah, so my my that fan that's coming from ⁓ Czechoslovakia, which I know maybe I'm canceled for that country name 'cause I know they're two different countries now. But I actually don't know which if it's Czech Republic or Slovakia. I don't know like which I've asked. There's this weird thing in America where like people like immigrants, like they just don't wanna or at least in my family, they don't want to talk about it. So ⁓ and that's like so I don't know, but somewhere from either Czech or ⁓ or Slovakia. I don't

speaker-2: This is gonna like I'm I'm Romanian, so I'm close enough. I'm giving you the pass, so it's okay. You're not gonna get you're not gonna get so

speaker-1: I was bracing myself for the emails already.

speaker-0: Is this a club not a part of the Eastern European club?

speaker-2: I mean honestly you could pa you could easily pass. Like I was thinking about this. Like if if you went to like Eastern Europe and you said your name was like Ivan or like Bogdan or something, I think most people will go like yeah, yeah, that's that's about right. I think you could easily pass.

speaker-1: Two.

speaker-0: Fair, like I'm extremely racially ambiguous. I've gotten everything from like Filipino to Indian. I am just the South Asian man morph.

speaker-2: I don't know. I s I saw you in like that ⁓ clip from Japan when you in your video. And do you do do you have green eyes or was that just like a like a reflection? Okay, yeah, so there you go.

speaker-0: No, I do have three days, yeah. One of those It was just the most ⁓ random Minecraft seed of all time.

speaker-2: Yeah, like I like I I saw you and I thought 'cause obviously I'd already I already knew what you looked like 'cause you you've been on the on the broadcast before before you were in the video. But ⁓ I thought, yeah, that's just a that it could he'd be my neighbour back in back in Romania, like yeah, no easily.

speaker-0: Interesting. A new a new contract.

speaker-2: Just say Romania. Just say I know Romanian he's allowed me to say, it's fine.

speaker-0: Brett, I really wanted to ask you. So from a technical perspective, what's your background? And what are sort of I've but I video on AI coding, by the way. I wanted to understand what your background is in software engineering, because I've noticed that a lot of software engineers have very differing opinions when it comes to AI coding. So I've noticed a lot of vitriol from people that wrote, say, C code and game development code versus web developers who have like

speaker-1: Dear.

speaker-0: I don't know how many agents is that now, like thirty agents, forty agents.

speaker-1: When I was 13, I'm in my mid-30s now, I started making websites, anime fan sites. So like I had a Dragon Ball Z fan website, I had a Trigun fan site, which Trigun kind of faded out of popularity these days, but I love Trigun. And ⁓ so I had these fan sites on GeoCities and you like would have to do a little bit of HTML. And then I had a fan, I had a blog where I would review episodes of anime, which is sort of this like funny thing to think about in hindsight. Like Like I would watch like an episode of Full Metal Alchemist and I'd like, This is a great episode. Like Roy Mustang used this fire. I'm like, you know, you're like, you got this WordPress site and you have to like put sick pictures of your Gundam like in your header. So I'm like learning CSS and HTML and like a little bit of PHP. And then my high school offered a Java computer science class. So when I was 16, I took a Java class and the teacher would print out the code and read the code on paper to grade it, which is awesome. I don't think he ever compiled it. So who knows if it worked, but we had to write a bunch of programs. It was all ⁓ programming and I made a game. I have that game still. I have the dot jar and I have screenshots of it. And it's like you pilot this Gundam or you pilot Moltres from Pokemon and these Gundams attack you. And ⁓ I got an A, so that was cool. And then after that class actually, I went to a summer sleepaway camp for game programming. Just like so weird thinking about these things. Like, why did this exist, you know, almost 20 years ago? And it did. And this was when the Xbox 360 was really big. And Microsoft had this framework called XNA. And then now there's an open source version version of that called FNA. So like games like Fez, games like Celeste, these are games that were made in largely a similar and there's also this fork of it called mono game, anyway. But we would use XNA and we would program games in Visual Studio. Was it called that? V before VS Code. Like there's like a closed source version before VS Code that Microsoft offered. They make they might still do. And you would compile your C sharp and you could run it on the Xbox 360, which was like mind blowing. So I went to the sleepaway camp and I'm like doing game programming. And then I went to school for game programming. But as I was in school, you're like paying forty thousand dollars a year or some crazy, crazy amount of money for school. And I started working doing apprenticeships, doing Ruby on Rails. And other programming, and was like, I'm making money. I'm not gonna pay money. I could be making money. So I left school and started working. And that set me off on this trajectory of just like web application programming. So like building web apps, APIs, databases, doing a lot of Rails and Ruby. Now, more recently, TypeScript. And ⁓ but during that whole time, which maybe I have regrets, like maybe I wouldn't make that same decision that I made. I wouldn't make I don't regret not being in s student debt, but I do regret not sticking with lower-level programming. Like when I left school, we were just digging into data structures and algorithms at like a more complex level and like doing C, and I left it to basically do Ruby. And I think that really changed my trajectory, but I still in my free time kept programming games. So I have this like weird mix mash where it's like I write a lot of TypeScript, Ruby. And Rust and lately a lot of Lua. And I I like messing with things. And I'm not good at I'm not good at spending five years on something. I'm like, I can spend like a couple months and ⁓ then I'm ready to move on. So yeah, that's kind of my mix, my mix mash is like web and some low-level programming. And maybe that's why I'm seeing both of those worlds, right? I'm seeing the sort of like web AI usage where it's just like fullbore. Like I've given it all, I've given it all over. And then you have the game dev space. Where you have art mixed in too. And like artists hate AI. Like I just you cannot, if you don't talk to art people or creative people, you don't realize like they hate AI for good, good reason, right? And I think that's something that as I have one leg in this game dev space and realizing, ⁓ wow, like people hate this stuff. And let me learn more about that. And yeah, I think it's really, really different and different perceptions where I think in when you're in like JavaScript land, you're like a next, like if you go to like the Next.js website. Like the Dino website. The websites now say have an agent install Dino for you. Just copy this prompt. And you're like, this is we've gone off the rails. Like if if we can't even install runtimes.

speaker-0: That's such an interesting background. I love it. Though that's like one of the most unique backgrounds I've ever heard. Like a game programmer turned web programmer, turned college dropout.

speaker-1: I think the game dev is industry is very difficult. So like in that regard, I all you hear is horror stories about like huge layoffs. Like I think within the last month or two, Microsoft laid off sixteen hundred people, laying off another sixteen hundred people. Like it seems like a very hard life to work as a game developer. So I think in that way I've also I'm like pampered by like web web dev or whatever.

speaker-2: Speaking of Web Dev and Rails, ⁓ because I'm curious, just to provide a little bit of of context, I'm a recreational programmer, I've never worked as a programmer just doing recreationally. And the first programming job that I ever applied to was the 37 Signals, ⁓ DHH's company for those who don't know, who also made ⁓ Ruby on Rails. And this was their junior position like a year ago, where there's like a super good position, 3,000 people applied, something like that. And I was one of the three percent who got a take-home exercise, just just bragging for a second there. Just for context, right? I know, I know. And I remember I was I was reading 'cause I knew about the company before. I'd I'd listened to their podcasts, I'd read ⁓ both ⁓ Davis and Jason's ⁓ blogs. And I've seen DHH's opinion on ⁓ on AI change pretty dramatically over the past year. And recently I was using Omachi until fairly recently on my on my laptop on my other laptop, and I kinda stopped because he ⁓ DHH posted a saying that he He recently added 30,000 lines of AI code to Omachi. And he said that he read it he read all of it. And I just I don't know what to think of that. I wanna I wanna know your take on this.

speaker-1: Yeah. ⁓ my gosh. I have so many thoughts on that topic in particular because so when I started using Ruby on Rails, it was 2011. And I'm coming from using C and C sharp and Action Script 3, which was like Flash, if people remember Flash. And I worked at this company that did solar panel monitoring. So it was like time series data from solar panels. And it was all processed with Rails and stored in a database and I'm exposed to Rails. And then you follow along with this project Rails, and there are some really prominent members. And you can't like, if you're keeping up with Mar like versions of Rails and what's added, you're like following DHH and you're kind of following a trajectory. And I, you know, he has written books with his business partner Jason. I've read those books, and they had a really interesting take on work and how to structure a company and working remotely. And like it was. Yeah, it felt really cutting edge at the time in the early twenty tens and mid twenty tens. And then, like, as time has just gone on and on, and this was he was always really strongly opinionated on Twitter and like just kind of seeming to Yeah. And it was just like that that's just not how my personality is. Like, I don't really want to keep up with you. But more interestingly about all of that is like, I don't think you're reading 30,000 lines of code. I I've that AI spits out. You're like maybe he's reading the key bits. ⁓ but I go that's that's really hard to do. ⁓ I've been on the other side of that. My one I ha have an open source game engine called Usagi Engine. And it's like this 2D game engine that lets you live reload your code, almost like HMR and web dev, so you can see the changes right as you're changing it. And it's open source and it's public domain. And the first PR I got was this person using OpenClaw to to add they they they wrote ⁓ an an abstract syntax tree parser for a custom ⁓ shading language that took the code base size from ⁓ like 10,000 lines or something like of Rust to 30,000 lines of Rust. So this person submitted this 20,000 line PR. It was the first PR of the project. So I was like, ⁓ that's exciting and I love that this person submitted it. ⁓ but it was like, I can't maintain this. This is all slop. There's like it's not understood or reviewed, and I have to maintain it. So just saying like reviewing that much code is Very, very difficult. Code review is one of the hardest parts of coding. And I think when you shift your job to being I'm just a code reviewer and I do that eight hours a day because the AI is outputting, I think that's really dangerous. But really quick, back on Rails. This really interesting thing happens where Ruby is this like beautiful fun language. I think it's like really good for scripting things. Like if you have like a folder of MP3s and you want to like mass update the metadata, because I still have folders of MP3s because I'm just like a weirdo who uses an MP3 player. Like

speaker-2: All legally owned of course, with all the rights and everything. Yes, yes, yes.

speaker-1: Absolutely. I have so many CDs and I ripped them all. And that's not even a joke. Ruby is so nice for that. And it's a great language for more than just web dev. But because Rails and this really strong personality of DHH took over or leads the pack there, it really, I think, suppressed Ruby as a language. And I think it really it I think it's bad for a language and bad for a community to have like one all-star project that dominates the conversation. And then you have these companies like GitHub and Shopify and 37 signals who then dominate the space. And I think that that's unfortunate. And there are cool projects that push against that and and really neat things happening in the Ruby world. But it's just ⁓ I think it's really dangerous when you have one figurehead person at the front of anything. I think you see that too with Linus Trevolds. And I think you see that with him talking about how AI is quote unquote just a tool. And that always worries me. There are two things that worry me in in open source, especially now with AI stuff, is like you have. One person with a really strong abrasive personality running something, and you see that over and over. Or you have like VC backed open source projects like Dino and Bun. And both of those are unhealthy in different ways and make me so concerned and nervous about choosing technologies or like what you use or don't use. And I'm off I'm off in the weeds now, but that's yeah.

speaker-0: Yeah, no, dude, like that thing you said about a technology just being overshone by a personality, that's like most technologies. I I feel like in modern discourse, this is gonna sound this is a very normy take, but it's sort of like how the primogen got associated with NeoVim. It's not that he is, say, the biggest contributor to NeoVim. It's just he's the one that popularized it. And now if you go on YouTube and you're doing some cool s cool stuff with NeoVim, prime's sort of the marker for that. And That's sort of the same thing that happened like early days with Ruby. I and I also agree, Ruby is a beautiful language. I've used it personally f even without Rails to do like very small script. And it has that sort of like robustness of Go, but it's a lot more expressive. Whereas Go, you're kind of like if you've ever read like a tricky line of Go, you're kind of like, yo, chat, I have no idea what this means. But with Ruby, it's like the language does not allow you to do that. It's a very expressive language. And also on code review. I'm not reading AI code. I'm just not. Like I am not reading AI code. I'll get an AI to review it and do it. Because the way I used to do code review professionally, and this is if if we have a junior developer listening, this is how I'd like everyone to do ⁓ code review. So initially when I was like a junior dev, maybe five years ago, I would like actually sit on GitHub and read everything. Whereas like I had this mentor and he came on, he came on call and he's like, dude, why don't you just clone the branch and just diff the changes and do like read locally? And that way, this is where this is where commit messages actually mean something. So if you start reading a whole pull request from the first commit in your local environment, I used IntelliJ because I worked at a Java shop, but you could do this with any tool. But if you start reading everything from the first commit, that's so much more powerful in doing code review because you can be like, hey, yo, your commit here was actually kind of better than your final commit. Do you want to actually like swap this over, maybe rework that? It's just so much more valued. in understanding a person's thought loop. But an with an AI, with AI code, you're just building a black box. I don't care about your black box. I've never looked inside my Mac. I'm not gonna look inside your AI code.

speaker-2: Having someone being the face of a technology. ⁓ So obviously Steve Jobs was the primary face behind Apple. Not just behind Apple, but behind the idea of building beautiful machines. There was this sort of dark period in tech where very early on machines, like if you look at this stuff that Olivetti, who initially made typewriters, and they they built like the well, the first like programmable calculator, like the programmer 101, I believe, which was this beautiful like machine. And then there was this dark age of computers look kind of ugly. And then you could argue that ⁓ Apple came in and sort of started building some beautiful looking machines again. ⁓ obviously all of that was driven by one guy who was at the forefront of this. ⁓ do you agree with that? Do you think that's that's different compared to having someone like DHH ⁓ as the as the face, essentially the face of modern Ruby?

speaker-1: I I think there's something interesting that happens when you have a person drive something forward, especially if they have a vision. And I think there's the problem is though, is if that becomes the person or the personality for over a decade, and then that person becomes what we would say today problematic. Like I don't know if the person who made that program ⁓ type like calculator is like ⁓ a horrible person. He might be, right? Like, who knows? It's tough to say, but like The problem with it now is like you can have someone who has a vision and like drives something forward or innovates. And that like typically I think if you ran that by committee or you had this like from the genesis of a technology, like a new open source project that you had it immediately driven by committee or some sort of democratic process from the beginning, I think that's kind of hard to ⁓ try things out and just run with it and see what happens. So it's like, I think it's good. when you have a person who sets something up and establishes it. But then when it's just like, ⁓ you're 20 years into this person being a quote unquote benevolent dictator, which again, if you remove benevolent and then you just have dictator, you're like that's actually no one wants that, I don't think. ⁓ so it's like these light these projects have these life cycle where they stop innovating, right? Like they do the innovation and then now they're just like stable and boring. They're almost like enterprise software. Like you could almost think of Rails as like boring enterprise framework. And it's not innovating like it used to, and that's okay. I think that's fine. I think you have businesses that run on it and they get the security patches and maybe they get a couple new features with each release. But like I think that the way that s open source evolves and changes, I don't think you just want one person. Once you get to that boring, like like Linux, like the kernel, is that like I don't think you want just one person dominating that that thing. I think that's ⁓ bad for the for the health of the project in the lottery.

speaker-0: Anyone should dominate a technology. That's like I'm a firm believer of like not having one person be dominated with the technology. That's why in a weird way I liked that Bond sort of went off of Zig. And now Zig can sort of okay, you can talk we can talk about Andrew Kelly's contributions, but I don't think Andrew Kelly actually largely cares what's done with Zig. So that's a bonus. Whereas stuff like Apple, Ruby on Rails, kinda next.js. Maybe Next.js is in this bracket as well, because it's kind of it's just taken over the web dev space. Everything is Next.js now. I think they're yeah. You don't want anything to be dominated by a single person. That's sort of why Rust is doing so well. Yeah.

speaker-1: I don't know who makes next either, which I think is really interesting because I think JavaScript world agree. I you don't want one dominant player. I think Next is the perfect example of that. Like at work, we use Next and we have a Rails API. And some of the Next code calls server-side, you know, calls out to SQL. And Next.js until the most recent version, 16.3. The dev server used three gigs of RAM. So if you have multiple Next apps running, it like you can't, you, you hit you max out your RAM very quickly. Well, they recently made it use like 80% less RAM or some wild thing. And but the thing is, it's like, why did that take so long? How is this? How can you be the most popular web framework and have such a poor developer experience where it's like crashing locally all the time? And I think it's because you once you get, once you're like the big dog, you're like, well, whatever. No one's like, and I I kind of think of next as like a Vercell project. So rather than it being run by an individual, it's like this company. And that's this cord sort of other open source thing where it's like, I think Bun is also similar, where you have the guy who runs Bun, whose name I don't know, who like is doing all these stunts, right? Like, ⁓ I had AI, rewrite it in Rust and like all this stuff. It's like, I actually don't want to think about that person too much. And I don't think you want that person to be the face of your runtime. Like, I don't think you want to be using a runtime that is doing stunts. Especially AI stunts. I think that is not what you want to build a business on or a project on or whatever. And I think that's that's where it gets really ⁓ dicey and concerning for me. And I so I've been using Dino, which is an alternative runtime, much like Bun is, right? You have Bun, you have Node, and you have Dino. Bun and Dino's big thing is that you can run TypeScript and check the TypeScript without having to install the TS package and worry about maintaining that and configuration. Which is really nice. So I started using Dino about three or four years ago. And I love it. Like I use their web framework called Fresh. I use their hosting for my own little projects. I use it to write a bunch of little CLIs. I use it to build pretty much anything when I want to make something with TypeScript. I use Dino. And I really like the developer experience. But The interesting thing about it, so it's made by one of the people who started Node.js. Yeah, yeah. And he sometimes comes through and he sometimes has horrible takes, which is annoying, but okay, let's like put that to the side. I just want to think about the runtime. The problem with the Dino runtime though is that it's VC backed, so they got twenty million dollars from whatever, right? Some number of years ago, and then they had twenty or more engineers, and that money is burning and they're not able to make money. So they like have their hosting platform and they're doing like enterprise stuff. And n that now they like it. They're trying to get money and they're trying to figure out how to build a business or get these investors their money back, but you actually don't want your runtime maintained primarily by people who are trying to figure out how to make rich people richer. That is not how you like run a healthy runtime, I don't think. I think you want a boring runtime. Like Ruby is actually really interesting, where you have Matt's, who he doesn't work much on C Ruby anymore, but you have this kind of core team of people, and you don't think about it. There's not much drama. They make it faster, they make it stabler, they fix security issues. That's what you want. But instead, what you have with the Dino company right now is ⁓ you have Bun, who gets acquired, right, by Anthropic, because Anthropic and a couple of other coding harness CLIs are built with TypeScript using Bun. So ⁓ the Bun runtime compiles it and they have an executable, executable. That's a weird internet joke saying it wrong, sorry. But you have this runtime that gets bought by these AI psychos, right? And now, well, I don't want to touch bun. I don't want to touch bun. I d I d for for a variety of reasons, but also I really like Dino. And I like that Dino has linting and formatting and this sort of like really nice developer experience. But then you kind of like sit there and you're a little worried because you got this AIR on their blog. They just had either layoffs or mass exodus of the main like main core members of the Dino runtime and their various software. ⁓ they redid their hosting platform and removed features and made it worse. And it's like Like what is happening here? You you know, like like I don't think you want I don't think you want a sort of like problematic maintainer who's the quote unquote benevolent dictator. And I don't think you want like venture capital backed open source. And I think that that's just yeah, it feels like a really weird time to have open source. And then you throw AI in the mix and it's just like

speaker-0: AI has fully ruined like any semblance of trust we had in open source. It's just gone. Like I I I thought people trying to contribute to open source before AI was bad. Like I would explain to people like you cannot just update README like 10 times or fix like a two line change. Like if you want to be a if you wanna be a good open source contributor, you wanna treat the project like it's your own project, build it locally, use the project. You know, it's like your piece of software. That's the point of open source. And then as time goes on, say you've set up security alerts, like there's a there's a tomato timer app in on Mac that I used to use when I was in uni. And I was so obsessed with it that I'd set up security alerts. And eventually I put that up as a pull request. And I feel like that's a natural way to get into open source compared to like get my open claw agent to spit out 30,000 lines of code.

speaker-1: And when you submitted that PR, you probably learned a ton. And that's the thing, like like the old school open source, and that still does exist. And I think that's like it does exist in these pockets, but these big, big projects make me sweaty. But I like when you submitted that PR, you have to read someone else's code and understand it and contribute to it. And that's like that's like what doing software development is, right? Unless you're like write everything from scratch.

speaker-0: It's the closest to the job, like inheriting someone's code base, realizing that a 19-year-old intern wrote all the core infrastructure. And this is why we have like four outages a week. I'd say open source, if you do it the old school way, if you fall in love with the project, that's the only way you can actually replicate a software engineering job. And VC, I don't know. It's like, have you guys ever seen the notion versus obsidian memes, like building software for VC versus building software for people? Obsidian only has three engineers. And they keep running laps around Notion. The same thing with like Lee Chess. I don't know if you guys play chess, but Lee Chess is one engineer versus chess.com that has 150. Guess one guess which one's VC backed? So we're sort of seeing this era where consumers are hyper-aware that software made for people versus software made for valuations are two very different things. And and the people that get it the most are programmers because we just know the VC world so much better than the average normie. Like if I go up to someone and be like, Hey, did you know your your like hat is owned by a VC firm in Denmark? They'd be like, You want me to call the cops, bro? You you too close to my face, man.

speaker-1: Like

speaker-0: They wouldn't care. Whereas like in the programming community, it's like it's kind of deep knowing that like, okay, this is like, for example, what Andrew Kelly does with Zig, right? The way he breaks down the costs on Zig, I'm not sure if you guys have ever seen his cost breakdown page, but it's beautiful. Like you can see how much he cares. And I don't think it was right of him to crash out at the Bond Creator the way he did. It was very funny, but I I don't approve of that behavior. But he cares about his project. And that's what's missing in programming right now. People caring about things. Like slop existed before, don't get me wrong. Like I've seen slop TypeScript. I've seen slop Java. Java is slop, but people caring is what's missing. And you have so many people now LARPing as programmers. Like they come in, they open a GitHub, they have like a hundred markdown files and like the most god awful projects I've ever seen. Whereas like there's still pockets, very small, like dying pockets of the internet. where caring about a project and putting in a lot of effort is still valued and you need to find those pockets wherever you can.

speaker-1: Absolutely. You said something too that I think a lot about, which is the idea of like who we're building the software for. And that's something I'm coming back to a lot lately, which is like, are you building software for other humans? Because when you're building something for other humans, you prioritize different things. When you're building something for a venture capital firm or investors, you're you prioritize different things. And when you build something for AI agents, you're prioritizing a very different thing. And something I have noticed too is like from Cloudflare and Vercell and Dino and all these different projects, is they've actually shifted towards building things. For agents and for AI companies, right? And I think that that is just I couldn't be less interested in it. And it's it's an instant turnoff when I read anything like that because like I wanna build I think the best parts of software is when you're building something for other human beings, either other developers and it makes their developer experience better or lets them build something faster because you built a library that does a thing that you don't want to do, or building something for a human that makes their life either better or more joyful or whatever it is. And I think when you do that by hand and you're thinking about the human, even if you're a business and you're trying to make money from that person, if you're doing it in a way that is ⁓ respectful and not overly exploitative, like that's like a connection, right? And I think that connection is really special. And I think when you use agents to write code or you're building things for agents, you lose that. You lose that humanity. And I think that that's like the the best part of open source for me growing up. In this career, growing into this career basically at the launch of GitHub. Like when I had that first job doing Rails, GitHub was like, it was like the coolest thing in the world, right? Like everyone had GitHub stickers and GitHub hoodies. And it was like it felt cool to be a coder, which like in the in hindsight feels a little cringe, but like it felt so cool to be a coder in 2010, 11, 12, 13. And then now you're like looking at GitHub and you're like, ⁓ you're just Microsoft. Like You're just Microsoft wearing like a Rails suit and you can't keep yourself up because you're down every goddamn day and and because you're you're you've gone all AI and you c you have infinite money and you can't stay up and and you realize like, ⁓ yeah, this actually like the cool thing to me about open source or what felt cool at that time, which also in hindsight, why wasn't anyone like, Why are we putting all of our open source code on this closed source platform? That's literally bonkers, right? Like Like that makes no sense. But the thing is they had this great user experience and they had pull requests, which was like hugely innovative at the time because otherwise you're like emailing patches back and forth. So it was really innovative. But also it's like, why was no one like, why is this closed source? We should be the whole ethos of this company is like fraudulent in a lot of ways. And now it's like coming back and and reflecting on this a lot recently is like, I think the coolest parts of open source are like. I think it's super punk rock. And I don't not everyone like thinks like being punk is cool, but like growing up, that was like the cool thing was like to be like against the establishment, against capitalism, and to basically ⁓ like open source has these tinges of being really cool. And I think that is appealing. And I think the connection with humanity is really appealing. And I think that's something that gets totally lost in this modern era of like why open source is fun. And I think The moment you take humans out of it, what are we doing here? So that's like what I'm trying to I'm thinking a lot about and I'm trying to like promote, I guess, is this idea of like collaborating, talking, making things by hand for other humans. And I think that's I hope we don't lose it. So

speaker-0: I just wanted to add, remember that conversation we had about like a certain technology being dominated by a person? People don't realize Git's a separate technology from GitHub. There you know, you like I would recommend anyone that is moderately interested in software engineering. Everyone at least has a 10-year-old laptop. You could run Git T on this. You know, I think th there's a there's a software engineering renaissance of where we have to go back to the basics. We have to go back to self-hosting. We have to go back to handwritten code. We need these things. These things are very important.

speaker-1: I use I host my open source code on Codeburg, which I think is a really interesting project. I think Speak, you might be interested in this too. It's run by a nonprofit in Germany. It's member run, which is really cool. And it uses Forjo, which is basically like a fork of Git T. And it's open source. It's run through a process. So it's not run by some venture capital company or Microsoft or big tech. And it's source code hosting and collaboration. ⁓ and it's a it's a wonderful place. Like if you're feeling like, ⁓ I I want to get away from GitHub, or like, like for me, I don't use any Google products. And I think that's something where if you're like, I actually want to get away from GitHub, there's options. And there are community members and people submit PRs and issues. And this interesting thing happened. So I mentioned that that person submitted that 20,000 line Rust PR for my little game engine. Well, when you actually move away from GitHub and you move away from that space, people aren't, people aren't going to go and sign up for a A Codeberg account and submit a pull request 20,000 lines with open claw. Like it it actually like protects the project a little bit. And and you it ends up feeling closer and and like the people who are committed to the project, because there's just like a little bit of friction, the people who are committed to the project are regulars and it feels like a community. And I really it feels a little bit like GitHub used to feel in the old days. And Codeberg even recently banned. AI projects and like really taking strong steps to have their own space in a center of the human. And I think that's a really, really important, especially in this era as like GitHub ⁓ automatically has copilot on your open source PR reviews and it's just there sitting there and you like can't turn off the agents tab and like it's just ⁓ yeah, I think GitHub's like totally cooked, man. Like Like with the outages and the UI experience and like the priorities. Because you know someone at Microsoft, there's some exec in a suit who's just like, You gotta be shoving AI down their throats. Like we are so behind. How are we so behind? We have all this other open source code in the world. How are we behind on this? Like, why are we not top dog? And you know, it's who knows why? I don't care why, but it's just we'll only continue to lean harder and harder into that.

speaker-2: Curious, ⁓ so I'm hearing a few different things from you guys, and I I wanna the combination of these ideas might be interesting. So BigBox, we said that we need more people who care about ⁓ the projects that they that they that they make. ⁓ I'm thinking about ⁓ the original Linux, maybe FFMPEG, like people are making like passion projects, stuff that they're really interested in. But then the impression that I get, Brett, from some of the things that you're saying, it almost sounds like it's possible for a project. become so big and influential that it should be treated almost like a public utility. So like FFmpeg, Linux, maybe they should become public utilities if they reach a certain point. ⁓ what are your guys' opinions on this?

speaker-1: Yeah, I super agree. I it's like almost like a national treasure or something where you're like, okay, this r there's enough there can still be someone leading something, but it should be there should be multiple people in the room. I don't know how Linux is developed, honestly. Like I don't know if it's just Linus decides AI is okay or if there's some discussion. Like I actually I have no idea. But I do think when something matures or stabilizes enough, I think it's a good thing to have multiple voices and transparency. And I think that's actually something Rust does really well. So like Rust Rust respond out of like Mozilla and it's really transparent. And there are meeting notes. Another good project is Bevi. Bevi is younger than Rust, but it's a game engine built on Rust, B-E-V-Y Bevi. And they actually keep their meeting notes public, the organization is public. The pay of the people who work for the Bevi company or foundation, whatever it's called, is transparent. And you can like follow in Discord or on their website. And I actually think of all the projects I've seen, and I don't follow this stuff super closely, so I'm sure there's other Companies doing it or projects doing it in an interesting way. But I think Bevi is like and Rust are two really interesting projects where they are transparent. They're trying to do things differently. Bevy also banned AI really early on. ⁓ and we're like, we can't keep up with the PRs, and it's just like we can't maintain this. ⁓ I think that's a really, yeah. Interest though, there is a like, there is a person who like started Bevi. His name, ⁓ I think name is Chase. I could be wrong though. But the there was a person, but he like actually stepped back. And there's actually a few people who maintain it and are sort of the more public presence. And I think that's actually like the best counterexample now that you mentioned asked that question of like a person has a vision, builds a thing, sets up this community, fosters this community, and then realizes that, you know, there is we need to have more s voices in the room and a process for these things. And ⁓ yeah. So Yeah, I agree with what you said.

speaker-0: This is such a cool project. Sorry, this is a whole new rabbit hole.

speaker-1: The way Bevy is run is really cool. It is very interesting and good it's a good it's a good technology ⁓ but the way it's run is so fascinating and ⁓ yeah, big respect for it.

speaker-0: Yeah.

speaker-2: So in your video you took a stand on not using AI ⁓ at work. ⁓ has this impacted your work in any way? Do you feel like you're falling behind? Is there any downwards pressure for you to continue using it?

speaker-1: Yeah, no downwards pressure, which is good, and no there's been no repercussions. I think so we're going on three weeks now, and it's not like I like I was not using AI for my own personal programming. So I was like back in the groove since the beginning of the year. But then not using it at work, it's like At first it's like, huh, ⁓ I have to like use my brain. Or like PNPM output an error and I don't know what that means. And so, ⁓ okay, well let me actually read the message. That's a good start. Let me actually read it. Let me read the stack trace, which you stop doing when you just kind of go full AI. You're just like, I'll just paste it into the thing, whatever. So it's like, okay, I read it. And then I go and I search online and learn more about it. And and so I've learned much more. I've learned more in the past two weeks than I have in the last year, which is sort of embarrassing to say, but true. And then like each day is much better. And I I didn't anticipate that. Like I have to, I'm applying my brain more. And I think the human nature might be to like, I just want to like lie on the couch and like ⁓ do nothing because I think that's like human nature at some point. But then you realize that that's actually not what makes you happy. What makes what makes me happy is doing something challenging and interesting and applying myself. So now I'm doing that every day. Rather than just like when people say things like, ⁓ you're a manager of software and of these agents, or you're like an architect and you're like telling them how to build things. The thing is, though, is like that isn't enjoyable for me. I can't speak for anyone else, but like I've tried, I tried multitasking with the agents, tried to do all these things, and it was mentally so unenjoyable. But then to say, okay, I'm building this by hand, I'm building this feature, fixing this bug, it's significantly more enjoyable. It's more fun. My days are better. And and there's also this other little thing I didn't bring up in the video, but I noticed this pattern, which is like one more prompt. Here, I well, let me I'm just let me just do one more ticket quick. Well, just one more prompt. And then you're like, ⁓ wow, I've been working for 10 hours. I feel horrible. Sure, maybe I checked some boxes, but like I don't actually think I did anything interesting or high quality. And I think that that's when you step back from that and you have a more natural, like, The way software has been built for decades, this cadence of trying to do a thing, like I'm gonna build a feature, learn what code exists around it, build it, add tests, test it yourself in the browser or whatever the interface is, review it, ship it. Like there's a there's a there's a flow to that that when you use AI, it totally destroys that. And you end up spending more time like doing code review, which I think is actually really difficult and unenjoyable. I also think like because of the way the agents work, you get lazy. So you're like, I'm not actually gonna read these thousands of lines. And and no one else on the team's gonna read them, just like you said, big box. Like, you know, it's it's everybody's kidding themselves if you if you throw up a 10,000 line diff, like you're kidding yourself if you're gonna actually like, and especially if you have multiple team members, if you have six people on your team or 10 people or more, and everybody's throwing up these 10,000, 20,000 line PRs, like. No one's sitting there and reviewing them. Yeah. It's it's madness.

speaker-0: think another thing I want to present to people is that like AI keeps being shown as like this thing with a massive skill gap. It really isn't. Like it really is not. Like if you know like the really foundational software engineering topics, not even software engineering, like web development topics, like okay, it all starts from like an HTML route, you put it on a server and you click buttons. If you can understand that and you can code that, I'm pretty sure most people haven't even put a file on a server yet. That's that's a different conversation. There is no skill gap going on. It takes maybe like four, five-ish hours. And most of that is just waiting for your computer to download and set apps up. And then just finding your groove of AI tools, like what makes you like, whether it's like using a TUI or using those new GUI apps or using like the built-in editors. There is no inherent skill gap. It's not like, for example, the this like the difference between being good at C is understanding pointers. That is hard. That's because it's a hard thing. Being good with AI tools isn't hard. It's just very tedious. There is no logical there's no pattern forming in your brain. It's it's like, you know those people that love setting up their local dev environment and they spend hours like making sure like the pixel there's like a five-pixel padding from the left side of the screen. That's what

speaker-1: Exactly age next source.

speaker-0: Yeah, eight eight pixels. Yeah. That's what I mean. It's like it's like Unix pore. It's like

speaker-1: Sure. That's what I was thinking of when you said that. It's un Unix desktop. And like I don't mess with any of that stuff, but that's also I have two little kids. So like I just don't have time for that. Yeah. Yeah. Which is fun. Like that stuff is really fun. No, no shade for any of that stuff. But yeah, I I super agree. Is like ⁓ this is something that you said too, is like, okay, like let's let's say for a moment that you can go and you can use AI agent and you can go and say, I want you to make me a homebrew.

speaker-0: There's a time component.

speaker-1: piece of software for the Sony PlayStation Portable, a dead console, right? And it can do that. And it will spit out the code and it will do it faster than I can. I have no problem admitting that the thing is faster and that it has a wider knowledge or like an a a wider resource pool that it is pulling from. I've seen it, I've done it myself, right? But the problem or the question becomes or or what I experienced with that, there was this era last year, like fall of last year where I like, I am now software god. I can make anything, right? And that's after like 15 years of basically being hamstrung and being like, well, I can only make certain things and then I hit a wall and then I I'm stuck. ⁓ And so then you're like, ⁓ I can do anything, right? That's like this feeling I have. So I had I made like probably 20 or 30 different things, just weird, wild things like ⁓ internet radio application for the Mac or a game thing, right? Like all these different Things. And and this interesting thing happened is like, yes, it's faster. Yes, it is able to do languages I cannot do more quickly than I can. But the thing is I gave no, I didn't care at all about them. Totally disposable, right? Where if I look at the projects I've made in the past before AI and the projects that I make now without AI, I care deeply about them. I know the lines of code, I know the files, I know where they go, I care that people use them. I want people to enjoy using them. And it doesn't feel disposable to me. I think that is the big difference is. For me, as either an open source maintainer or even working for a company that is trying to make money, is like when you have someone who cares about something, they do better work, they know more, they're enjoying themselves more. And I think you can't undersell that. And I think when you only prioritize speed, which is basically what we're saying, right? Is like I could learn the PSP homebrew SDK and I could go and ⁓ maybe it would take me two or three times as long. But the thing is, is like that's probably time well spent. Cause I'll learn more, I'll grow, I'll understand things more, and that's also enjoyable for me. And then I care about it more. And I think that when you only factor in speed and time, which for some reason software engineers are just like deeply obsessed with, and I understand why businesses are like somewhat obsessed with it, but I think there's like this big fallacy here. And ⁓ that I wanna give make two points on. One is that for the last year, at where I work, let's say that, let's say generally, using AI, let's just say Three times more output. I think that's actually pretty accurate. Like if I look at the number of pull requests and commits and like linear tickets, probably like 3x more output, right? Has resulted in zero increased revenue. Zero. Now, is that because I'm building the wrong thing? Is it because that throwing more features or lines of code at a thing doesn't translate to revenue? Is it, you know, it could be anything, but it's not this magical thing that prints money. And I think that's everyone thinks it is. If you do more code, if you do it faster, it will either make you better or it'll make you more money or whatever. Not true. Now you actually have more things you have to maintain. which and the AI actually, the more code you have, the worse it is at maintaining it. So, okay, it's not like always this 100% gonna make you more money. Okay, well, that's interesting. Does it still useful? For me, I don't think it is useful. But then the second point is that if you take this tool, And you de-skill your entire team, and they don't know how to write code, they don't understand the code that's been written for the past year, it's not making the business more money, it's actually costing the business tens of thousands of dollars in tokens. ⁓ who's this helping? And why is speed the thing we care about? And let's say we have a paperclip factory. Because people, especially in YouTube comments, love to say this is the new industrial revolution. Okay, let's say let's talk industrial revolution, let's talk manufacturing. Let's say that you have a paperclip factory. Let's say speak that you own a paperclip factory in Romania and you are just like, I've got these people and I've got the demand and they are making paper clips. And sure, they are basically dehumanized. They get one break, if that, and maybe they're losing hands and fingers in the machinery, but all the time. But that's okay, because I am the business person, I make more money. And it turns out that there's just this huge demand for paper clips.

speaker-2: All the time.

speaker-1: And now I have a technology that lets me do three times more paper clips. And now speak, you are rolling in money, diving in that cash. That is like basic it like that works in the industrial revolution. But it does not work for software, which is like no one is talking about that. Which is like if you apply more productivity to software, it does not translate to making more money. And it means more work, more to maintain, you less, and then there's a huge cost to it, the emotional cost, the skill cost.

speaker-0: Means more work.

speaker-1: And that having three times more lines of code or features or whatever doesn't make better software. And that's this thing like, how much software do you use where you go, I wish this had three times more features? Like, no, you want software that's faster, that's more secure, that's more stable, and like generally is like out of your way. And I think that's my favorite software is usually it's like made by a small team or one person, like you were mentioning the chess app. Like, there's a connection there that you have with that person. And I think that that's Totally lost when you have management and they're thinking about productivity and output, but it's it's not a paperclip factory. There is no direct correlation between lines of code or linear tickets completed and necessarily revenue. Some places might have that, and that's good for them. But I'm just saying, generally, in the abstract, it is not as simple as. factory production processes. And I think that gets totally lost when we when all we prioritize is time, right? The time to get things done. Where really I'm interested in like, how do people feel? If a team feels good, they get way more done at a way higher quality. If a if a team is learning and growing, well now in two or three or four years, you have this like super team, right? And that's like a great place to work. I'm thinking about how how are we going to maintain this? How are we going to keep this secure? Why does it keep reading my ⁓ files? Why does it keep reading it? You know, you're just like, this thing is just inherently s insecure. And then ⁓ they'll say, ⁓ you gotta run it in a sandbox and you gotta who's doing that? I don't know anyone who's actually doing that, right? So it's just we're we're we're prioritizing the wrong things. And I think it's gonna have huge detrimental effects, especially then if you like bottom out junior engineers. Like if we're basically like not training anyone who would otherwise be learning the basics of, like you said, like pointers and C or ⁓ data structure and algorithms or how to write code basically without these AI tools. Like the if the whole premise is, ⁓ we have to have someone who can check the AI and take responsibility, well if you have someone who doesn't know when it's wrong, like what are we gonna do? Like it's like sorry, I'm like going I'm just ranting.

speaker-0: No, no, I totally agree. Like, I think an analogy I always come to is like software is a bit like food. Like, do you really want to idolize someone like Gordon Ramsey or the machine at McDonald's? Like, take your pick, bro. Like, I know whose burger I'm gonna eat. Both. I mean, both have their own utility. I'm gonna eat both. yeah, it's ⁓ software is also just like a it's such a deeply Personal thing. I don't know. Maybe it's because I'm a bit a bit of a nerd, but I deeply care about where my software comes from and how I feel using it. Like, sure, Notion can do functionally everything Obsidian can, except maybe like Vim key bindings. But there's just a coziness. Like software, software gives you a feeling. I don't know what it is. Maybe it's the designers, maybe it's a really good product manager that's like put this into my brain. But even when I'm using something like something as simple as a terminal. Like a terminal multi like a terminal multiplexer, right? There's so much thought and consideration, but if you're using a good one, you forget you're using it. I feel like these like so software, like the problem is we took, we idolized the fast food in software. We really just idolized the fast food. We idolized people like Mark Zuckerberg, Jeff Bezos, who weren't really even programmers. One of them was Laquant, a finance pro. One of them was like a really, really, really talented college student. When we really should have been idolizing guys like Fabrice Billard, like some dude in a French farm who just like wrote a library. For the fun of it. And now it runs this call, it runs YouTube, it runs everything. Or even like, even if you want to idolize someone rich, go idolize Mitchell Hashimoto. Like the first thing he did when he became a billionaire was to start an open source project. Like he started Ghostie.

speaker-1: Like maybe we shouldn't idolize guys. Like maybe that's just the thing we shouldn't be doing, right? Like maybe we should just abstractly be thinking, like, I want software that's made by humans for humans, and especially if it's open source, like I want there to be some at least transparency or some sort of like path towards stability, or like, like, let's stop the AI stunts, let's stop the VC backed stuff. Like, that's cool that Mitchell's like, you know, my my rich guy retirement isn't gonna make a T T Y. Like. Whatever, like that's cool, but like I I think we just

speaker-0: No, I'm I've just fallen into the pattern of idolizing people in software. So I'm just like, okay, if you're gonna have idols, if you're gonna have role models, which I also agree, I I completely reject the idea of like role models outside of maybe sports and music. ⁓ but if you're gonna have role models, idolize people that are computer scientists and not just goobers that can convince enough people in suits to give them money and then hijack our attention.

speaker-1: Yeah, and don't care about like don't be looking for takes. Find a mentor. Like if there's one thing that I'm just like consistently going back to is like you gotta connect with humans, right? And like that's why I was so excited to chat with you both, because it's like I wanna talk to people about this stuff. I don't want takes, I don't want stars, I don't want likes, I don't want numbers going up, right? That's like the biggest problem right now is like we're just want numbers to go up. I just wanna talk to people. Find a mentor, right?

speaker-0: This was my hottest take, like on Twitter once. And I was like, if you're if you've been programming, like if you're a junior developer with less than like five years of experience, remote work shouldn't be in your vocabulary. It just shouldn't. Like I have learned so much just going to an office, sitting down with my laptop and having like a senior engineer just watch me code, and then him going, like, all right, give me your laptop, bro. This is over. Like, let's let's sort this let's let's get go to a whiteboard, sort it out together. That is so much for my development.

speaker-1: Super agree. And so I started working, I was in that office when I had those apprenticeships. And then I started working remotely. And the only reason I was able to grow and have that experience is that my mentors at those shops, who I stay in touch with, we used TMUX and we we basically tmuxed into each other's machine and were running Vim. So everyone had the same Vim config. Everyone had the same TMUX config. So you could SSH into anyone's machine and pair program. And so This person was like a huge advocate for pair programming. And I would drive and forget how to type and not know, you know, HTTP or whatever and learn these things. And yeah. And that that person was my mentor, right? And I think that you gotta have the synchronous learning. And I don't think that talking to an AI agent is the same thing. You gotta, you gotta pair programming, especially in that like learning in those early days. those early years, so important. And doing it in person too, like, yeah, it's all about connection and humanity. I think that's we can't lose it.

speaker-0: I think people get into programming a lot of it is like for the money. You can you can not be good at programming and make a shit ton of money. There's a lot of people out there that know nothing about code and have made money from software. But if you actually want to be good at it, that's a whole separate ballgame. And whether or not you make money is a very different conversation.

speaker-1: Very true. And that's something very related because that is what the YouTube comments are basically like you are cooked. You are cooked on like you're dinosaur. If you're not using AI, you're done. Find a new career. Good luck in the nursing. Like good luck becoming a nurse or an electrician or whatever. I think that's something that like if that's your brain, like if that's the math you're doing, which is like I program, I get paid, I get it. There are lots of people who have jobs where they're not passionate about the craft and skill. Like, that's fine. I didn't start this to make money and I don't do it to make money. Like I code in my free time. like and I don't make money for it or whatever. And like I think that's something that craft, the skill, the enjoyment of it, that is like so special. And I think it shows in the software. And I think it shows ⁓ in the relationships you have. And yeah, I agree. And and no shade to people who do it just to make the money. Like I

speaker-0: I get it. You gotta pay the bills. There's no UBI yet.

speaker-1: This isn't Star Trek future. I don't see any hints of Star Trek future. I see hints of like huge economic collapse. Right? I don't see anyone, I don't see any of these AI psychos talking about UBI. I don't see them talking about like anything that will make humanity better. I see them talking about like pants am I allowed to curse? I see them talking about just like pants shittingly bad things, about losing jobs, like humanity ending, all of these horrible things because they want their company to seem like it's important.

speaker-2: You can yeah.

speaker-1: No one's talking about, you know, how this helps humanity. So

speaker-0: Here's one way, like I was talking to a friend of mine. ⁓ house prices are going down because of the data centers. That's great.

speaker-1: ⁓ great. I can't hear anything now because I have no clean water, sorry.

speaker-0: ⁓ Yeah. 'Cause the data center nuked the house price, the whole suburb's gone to shit. Like all the kids of primary school have like three heads now.

speaker-1: I was just driving this week. What do you see? Signs. Just signs everywhere. No AI data centers. I I generally don't think people like this stuff. I think there's just some like weird tech bubble where you're just in it and you're like, either my boss made me do it or I want to get more done, or I wanna like cosplay as a programmer. And you don't realize like, ⁓ yeah, like humanity hates this stuff.

speaker-0: Yeah, like I work at an AI company, I use AI tools. If I get an email that's AI generated, I am blocking you. Sorry, speak, go on.

speaker-2: ⁓ I was gonna say I think that there is generally a movement. So I think the AI has kind of irreparably harmed the way people look ⁓ view technology. I think that all this stuff started probably with the last industrial revolution. So if you look at ⁓ I was looking at I love documentaries about like the past, a few centuries back, especially the sixties, Palo Alto, beautiful. But I was looking at a documentary from like 19th century. And I noticed something like early early 19th century. And that accelerated with technology. So like humans used to live in houses. Now humans live in pods. Humans used to walk or ⁓ or ride their horse somewhere. Now they put you in a tube on the ground and they shoot you down a tunnel. Food used to be natural, like you'd slaughter the animal, or you'd pick the vegetables and the fruit and you'd eat it then and there it was fresh. I know you're vegetarian, Brad, sorry, that's why I sorry vegan. That's why Now the Now the food is highly processed, comes from across the ocean at tremendous ⁓ economic ⁓ and environmental cost, ⁓ more importantly. There's generally been a shift towards treating humans more and more on an industrial scale. And That was acceptable initially because of the economic benefits of it. Like our grandparents were like, okay, yeah, we're being treated a bit more inhumanely, but we can have like free houses and two cars, and it's great. And I think that as the economy, as the benefits of this start to dry up, as people get less and less economic benefits from having an AI data center near them, ⁓ or from just from technology in general, people are gonna become more and more. anti-technology. ⁓ that's why you're seeing more and more of a push for open source as opposed to using Mega Corps project. That's why you're seeing people putting ads saying no AI data center. That's why you're seeing Darth Vader ⁓ talking pro flock. If you've seen that video, it's ⁓ it's ⁓ it's brilliant. like I think that that this is we're just kind of approaching a combination of Humans being treated less and less humanely for less and less money and it's starting to make less and less sense.

speaker-0: This also reflects in like online culture. I'm not sure if you guys have noticed, but like the online sphere, and I'm not talking about our space in general. I think tech is still it's okay. If you guys go on like Instagram comments of like an interracial couple, it's crazy. Like some of the comments I'll be reading on like an interracial couple, like whether it's a ⁓

speaker-2: In what way?

speaker-0: A person of color with someone that's white or someone that's Asian with like someone that's a person of like a person of color. The comments are crazy, bro. So deeply, deeply racist. And like, I don't know if people aren't taught this anymore, but like it's not nice to hurt each other's feelings for no reason. No. Like, it's like maybe sometimes it's cool to be a good boy. I don't know. Hot take?

speaker-1: I think people are horrible. That's the hottest thing. It's so true. It's it's and I think speak, what you're saying about like you you have you you hit this threshold where you're like, this is not worth it anymore, right? And I think that's the thing I'm constantly thinking about. And like, I use this thing. I have a flip phone, a dumb flip phone. It's a horrible piece of technology. I had it for a year. I'm not just like making some video for a week. I've used it for a year. There's no going back. And I think the question always is when we use any type of technology, or let's just say, A tool, but let's say technology. The question is always at what cost, right? At what cost to myself and my mental health, to the environment, to my community, to my business or whatever, right? And I think that's this thing where if you just think you have to have a tech, you have to have a smartphone, you have to use AI to be a coder, you have to do these things. It's almost never true. You can almost live a better life without it, or or like social media, right? Like it's it's so possible, so true, or so possible to not use these things. And I think that's to to experiment, to explore, to ask questions, to see how it feels, to see how you feel, to see what changes and how do I use a thing, right? And what is the cost of it and what is the benefit of it? And I think that cost benefit analysis to use a a business boy term, it's really important. It's really, really important. And I think that the moment that you're just like just keeping up with trends, I think you're losing your agency. And we all have agency. And I think that's really So so important and heck yeah. Apple watch Apple watch I don't even I don't even have a watch because I don't know time. I don't let time run my life. You gotta

speaker-0: The watch just can eat my shorts. Like Garmin, put it in the bin, bro. This thing, fifteen bucks. I don't I don't need to count my steps, bro. I can just go for a 30 minute walk and it'll be twelve thirty. It's great.

speaker-1: Indestructible. And who are you sending your sleep data to? Like Yeah.

speaker-0: Exactly. No, you know how I know my sleep data? I feel great in the morning. Like or I'm tired. And it's like when people get all these like smart bulbs and stuff, like, bro, flicking a switch is not that hard. I'm begging you, flicking a switch is really, really not that hard.

speaker-1: Yeah, it is. hundred percent agree. And I think that's I don't I I think that's I don't get it. But I'm also like I think I'm on the fringe. Like I think that's the thing is like I'm like I'm in the corner with my flip phone saying AI is bad. And I think that everyone else is just like, you know, I talked about Dostoevsky with this thing. And you're telling me it's bad? Yeah. Like why don't you talk about Dostoevsky with me? Yeah. You know, that's the thing where it's like what what like what are we doing here? And I think that's what we're doing here is we're trying to make rich people rich. Right? Like we're trying to make rich people richer. That's what's actually happening here. Like big number, we want to see it go up. And I think that's you gotta always ask, like, who's this for? What cost is it to me?

speaker-0: I'm a person that's like a huge proponent of like proliferating programming. I think everyone in the world, it should be in school to know how to code the way you know what chemistry is, right? Like I'm a big believer in that. But AI has made me go like, nah, maybe we should have gatekeeped it. It's because some some of the slot I'm seeing from you goobers, like some of the stuff I'm seeing from you, like, holy crap, bro, like really, bro, G stacked. You you shot up a hundred thousand lines of code per day to make a blog. Like something that WordPress solved like twenty years ago. I don't know, bro. ⁓ MySpace solved it.

speaker-2: just I I think that this this sort of stuff robs robs you of a lot of different things like for example ⁓ so as a as a recreational programmer I j I get obsessed with projects and I just work on them until the obsession runs out right so ⁓ and typically comes back once in a while and I'll I'll become re-obsessed with the project for a few days. Like recently ⁓ someone someone in my comments I I may I made a video about I don't even know what it was, but someone mentioned Eliza in one of the comments, which is this chatbot from 1966. It was like the first chatbot. And I think the video was about how people people anthropomorphize AIs. And someone mentioned Eliza because even back then, people were using Eliza, and even though it just did some simple semantic tricks, people became enamored with if they thought it was intelligent. ⁓ my god, it's It's listening to me, it's speaking back at me. And so I thought, you know what, let me just make my own version just to pro I'm gonna make a video on this, but just to prove that AI isn't sentient, it's not a person, it doesn't have feelings. Like, here's a thousand lines of code where it essentially just looks for keywords. Right. So like it's literally just a program in C which looks for keywords and it has it. It generates pseudo random sort of responses based on the keywords that it receives. And then it has a fallback. If it doesn't see any keywords, it just does like interesting. Tell me more, or can you give me an example, right? It's a pretty simple technology. And I was thinking about it and I thought, what is it about this that makes it human? That what makes it feel human to for people to use? And it was the imperfections that made it made it feel more human, right? So like I I literally I wrote Two new functions which initializes her person her. See, I call it a her. I called it a her. Fuck me. Like this is exactly it, right? Like I called a program called Elise, I called it her. Even I wrote it and I know it's not a her, but I called it a her, right? Like it initializes her personality and it makes her make grammatical errors as it as it outputs the text. And then like stops for a second, it deletes the error and then it continues. And I noticed that that almost made it feel more human. Like, no matter how much we improve the responses, it's not as impressive as making it a bit more imperfect. And that's what actually makes it feel human. And I think that I think that the fact that AIs are fallible, I think that might actually make them feel more human. I think that that's why people keep anthropomorphizing them despite them being imperfect. I always I always I almost wonder if. If the AI was perfect, if the output was always correct, would people be less likely to view it as human and view it as a machine? Is the imperfection what makes people anthropomorphize it so much? Thoughts? Question mark?

speaker-0: Yeah. No, I agree. It's like everyone's waiting for that gotcha moment. Ha ha, liberal. Like with the AI, like, you know, like I I've done bamboozled you. Mm my take is like honestly, like 'cause I 'cause I work in it, like and like there's sort of like a culture around I I worked at an AI company. ⁓ we use AI tools. I've I can't ignore it. Like I've s I've unfortunately been in the position where I've had to gear a part of my life around that. Well, when I'm in my spare time, I'm like, how can I actively ignore this? Like how can I just like when I'm thinking about programming, how can I just go back to like it's twenty twenty two and I'm still on VS Code and like I'm living my best life with Stack Overflow? Like how can I go back to that? Like how can I just ignore this conversation?

speaker-1: Yeah, I think it's a and then it's trained on data, you know, it's trained on internet conversations and Reddit posts and you know, books and all this stuff. So it's gonna output text that sounds like what a human would say. And I think that's probably why people anthropomorphize it. And I think that's I actually think that's like a d a dangerous thing. Like I I think that's like sincere like genuinely like a very dangerous aspect of it. And I think I experienced that where I'm like, I'll just talk to you like a pal. And I I don't, I, I, I realize, ⁓ wow, this is actually bad for my mental health. And I'm sitting in the car, driving my kids home, listening to the radio. There's a radio program, and they're talking about people who there's this example I just think about all the time because I felt like I felt like, wow, that's wild. And I can relate to it. And it was this: this person was using the chat, AI chat, and the person. hadn't even graduated high school and was talking with the AI about their interest in mathematics and he was explaining some things he'd been thinking about regarding mathematics and the AI convinced him that he invented a new form of mathematics. And like for months he's, you know, in this one chat thread just talking about this new form of mathematics. And then one day he asks a question and it like bursts the bubble and it's like, I made all that up. You actually didn't do that. The star was crushed. Crushed. And there are there are that's like a s kind of silly example, but I can relate to how that feels. There are like more harmful examples too. And I think that's just one of these things. It's like, why? Why do we need this? Like, why, why is this the interface for this? Why is it chat? Why is it what are the harms of this? Which is like way too early to tell. Related is like, okay, for driving, you have to pass a test. And usually when you're of a certain age, you can drive and you you pass a test and That's there's like a gate. There's a literal gatekeeper. You can't just drive a car. I mean, you can, but you're like, it's it's like you're breaking the law. We know that cars are dangerous, right? Cars aren't just a tool. They're a complex piece of machinery that has a certain function, and we treat its safety appropriately. But with AI, there's like nothing. Nothing. Maybe the company that makes the thing says they're talking about safety, but also that's not how safety works generally. Generally you have like some outsider. who is ⁓ not invested in trying to make money, who maybe is an NGO who's really thinking about safety. And I don't there's like basically nothing. I'm not talking about like like ⁓ safety of like government secrets or hacking into systems. I mean like the safety of the people using it and the safety of the people impacted by using it. I think that's something too where you're just like, ⁓ yeah, this is this like wildly immature thing. We're talking with it like it's a human. ⁓ we're maybe getting attached to it. Maybe we're thinking we're making new forms of mathematics or maybe Yeah, we're driving to the ER at one AM, like and realizing like, ⁓ yeah, this is like we're just the test subjects. Like we're totally lab rats for these companies. And the earth is collateral. And have fun, you know? It's I d I yeah, I I don't know why the interface is human like chat. I think that's just bizarro. But maybe it's because it's they found it's the like best way to get people to use it. And I think related. Sorry that I just go off. If this is all horrible, just shadow ban me from the episode. ⁓ If you think about like search engines or YouTube or anything, Corey Doctor's insidification idea, right? Where you have this thing and it's good until it gets worse, and then they need to support the business class because they want to make money and then it gets significantly worse. Look at AI. You have these companies, they're subsidizing the cost of the tokens. They're not making any money, they're spending a ton of money. Let's say like one of them or two of them like exists 10, 20 years from now. There's no way Claude is gonna cost just 200 bucks a month or whatever it is. Like that cost is gonna go way up. The platforms are definitely gonna insidify. They have to figure out a way to make money, and it's not making silly AI Sora videos, right? Like it's this whole thing is ⁓ it's just like a big grand experiment, and like the whole world economy is tied up in it. And I think that's like it's irresponsible. I think it's dangerous. I don't want anything to do with it. So I think that's It's it's worrisome. Like I don't think this is its final form. And I actually something I've been wondering is in its final form or its more kind of stable form, let's say it's five times more expensive. Well, that's actually more expensive than an engineer. Like, or let's say it's ten times more expensive. And now you've de-skilled everyone. And then what where where are we left with, right? And I think that's something where I wouldn't go so hard on it. Like I think if you're like, I'm interested in it, maybe I have to use it. Like maybe someone, some Manager is sitting there looking at your token usage, which is like a horrible metric to measure. That's like the least useful metric to measure. ⁓ like, like think about think about the long game here. And don't like explore, are there ways to use it differently or not use it? Just know like this is not, this isn't it doesn't have to be this way. Its final form isn't clear yet. It's like Frieza. You haven't even seen my final form. Yeah. And I think that's That's that's enough cause for concern. So I'm worried about the safety. I am sincerely Yeah.

speaker-0: It's like Rajin Buu. It's like Rajin Buu. It just keeps coming back. It keeps coming back stronger and stronger. And you need a spirit bomb at the end to just like I'm sorry, I I I'm a huge Dragon Ball Zig that.

speaker-2: Yeah, we all like dragon balls in here, we're good.

speaker-1: Wait, can they call it Dragon Ball Z in other

speaker-2: Dragon Ball Z Yes, if you're in the UK it's Dragon Ball Z.

speaker-1: That's wild. ⁓ my gosh. ⁓

speaker-2: ⁓ so I think that the ⁓ the end game with the models, if it works out for them, is to just essentially just steal every company out there. So like if you're a business owner and you have a company and you're using AI for everything, for the customer service, for the coding, for the business decisions, everything, you're essentially just giving the AI company your secret source. You're giving them everything, including your customer list. So what's to prevent the AI company? Eventually, like they'll they'll keep milking you as they gather data, charging more and more as you get rid of more and more people, as you give away your leverage essentially, as the owner. ⁓ in the end, they'll just clone your company and just do just make a version of your own company for less money because they don't have to they because they get to run the tokens at cost, right? So they don't have to, they can do it for cheaper. I think that that's probably one of the better arguments for the business owners. I'm surprised that many of them aren't seeing this. 'Cause like, look, bro, if you have to deal with me bags, like with humans all day, at least you can you can you can say, okay, well, you can do this job, but I can find ten other people to do the same job for the same for like less money. So you can at least negotiate with people to to f to kind of keep wages somewhat reasonable. But if your everything is one company, if the company provides you with the labor, with everything. You can't negotiate. There is no and you can't play the two free AI companies against they they all know each other. They're not gonna like compete on price. They all know each other. And so you're essentially putting yourself in a position where you will have no power. You they will they will be able to charge you whatever they want. And so if you create and even even if it doesn't happen, you're still creating a s an issue with scarcity of skill. So if no one learns how to program anymore because toc because tokens, Then what's going to happen is you're going to be in situation where, okay, well, now we have to have agents for everyone, because no one earth how to program, and they control the cost of the agents of the tokens. So now your engineers can cost two to three times as much money because they can't code anymore and they need the the tools. And you are now completely beholdened to two to three AI companies that run the entire thing. It it seems so obvious to me and to many people that I talk to. Just not particularly to the to the business people who make who make these decisions.

speaker-1: And I think those people are worried they're gonna miss out, right? Like if that's what it is. You have big companies where it feels like, okay, we have to use it or whatever, like that's fine. But then I think you have this kind of middle tier, medium to small business where I feel like they're worried they're gonna miss out or be left behind. Cause that's what the the marketing is telling them, right? So like and then you tie up the world's economy in this stuff, and it's like, why is this all like this? It's to make rich people richer. Like it's not there's no like this isn't like some And they'll tell you that this is, especially when like open AI was more like non profit research, like, ⁓ this is for the betterment of humanity. How's this making humanity better? Give me one example. Give me one ex

speaker-0: AI cat videos.

speaker-1: They shut that down. So anymore. I'm open-minded to the idea of there being applications of this that maybe helps with medical things. Or maybe it helps but but even then, that doesn't then mean okay, it's de facto everyone should be using it and investing in it and you need it to use your business or whatever, like to run your business. It's it's a very it's it's it's

speaker-2: ⁓ Yeah, I know, I know, they're gone.

speaker-1: It's it's like this weird thing, like it's this hype. It's ⁓ hearing about it. It's also like kind of like new technology's exciting. Like this is a feeling I had when I started using it. It's like, ⁓ yeah, this is like kind of neat and like kinda interesting and like, huh, like, wow, I could make an I make it make a Pomodoro app or whatever, right? Like you you you feel like, ⁓ I'm I'm able to do these things. It's ⁓ I'm empowered. But I think the thing that I've come to realize is it's like sh. It's actually disempowering. And and you're consolidating power into these companies' hands. And I think that that's a really dangerous spot to be. And we saw it happen with Uber. We've seen it happen with other companies where the prices get hiked up. And I think that would really change the mathematics that companies that are relying on this or investing in it are using. And I I don't think you want to put all your eggs in that one basket. Like I think it's I think it's dangerous.

speaker-0: Also I I think about it from like a bottom-up approach as well. Like I often talk to like like I'm a huge car guy, like outside of like programming and stuff. I love I love my cars. And I'm specifically like super into like Japanese drift cars. And I'm sure you guys have seen like drifting, right? Like it is actually the most inefficient way to turn around a corner. The most efficient way is actually to slow, slow down and then turn and then accelerate. But it's beautiful. Like excellence. needs no justification. I'm not gonna go and question, I don't even know anymore. Nolan's there's no there's no more excellence anymore, bro. It's done. I'm not gonna go question how like why Christopher Nolan chose a 70 millimeter for the Odyssey. I'm not gonna question his process. Excellence needs no justification. This AI stuff is super, super easy to learn. If you're a very good programmer, it's a weekend. It's a weekend and like 500 bucks at this point. Like I'm not even gonna lie. It's like if you actually want to mingle with everything, like Like I hate Elon, but the Grok models are objectively quite good for the price you pay. Like for 30 bucks, it's like the it's like the Burger King. It's not as good as like the open AI stuff. So it's not McDonald's, but it's Burger King. It's a few hundred dollars and it's a weekend of your time. But you shouldn't like invest all your chips and getting good at that because there is no good at AI.

speaker-1: There is no good. There is no skill. And that is something that I I that that really gets me gone.

speaker-0: Like I'm not gonna hire someone 'cause they're good at AI.

speaker-1: People in my comments are telling me it's a skill issue. You're telling me I don't know how to prompt. You're you're cooked. Like you have no idea. There to learn if you have any modicum of engineering skill, to be able to have the AI output a thing or not. There there's not like this like, ⁓ I'm using TMUX with this other thing. Like like that's all just toys. Like the actual core thing of what you do with the AI is not difficult to learn. And I think that's a bad thing. Like I think you want to be learning and growing and trying things and like Asking the AI like what framework you should use or whatever, that's like not doing research. Like doing like give an example. Like let's say you are want to build a new app. And let's say that you you're gonna use TypeScript. And you go and you ask Claude or whatever, and you say, which framework should I use? And it can give you like a table of the pros and cons and tell you all these things. But like you actually don't know anything about that stuff. And maybe that's you just want to make a decision and move on and build a thing. But actually, there's like something immensely enjoyable and valuable about setting up a little app in all these different technologies in Nuxt, terrible name, in Svelp, in Next, in Vanilla JavaScript or Preact or whatever you want to use. And I think that that's that's doing the research. And I think that that's something where you learn and you grow and you understand the landscape more. And I think that when you remove all of that, like you you aren't growing. You have no skills. Prompting is not like There's not like a secret prompt to do the thing right or better. Like maybe it goes a little faster or the results are a little bit better, but like that's generally like having used it for 18 months every day at work. Like that wasn't it wasn't these tools and how to use them. That that was actually like you catch on to that really quick. And I think that that's that skill rot, that's the dangerous part. When you are on the other side of 18 months and you're like, I actually forget Vim or whatever. Like that that to me is Yeah, I don't understand I don't understand the obsession with the different models and the tool. But like I never kept up with any of that stuff. But I actually think if you step away, like if you if you don't drink from the fire hose of AI, and you just step back and you're like, okay, ⁓ maybe you're using it as a tool for your job, you kind of realize like all this stuff is like it's not interesting. It's

speaker-0: It's a bit like the tech review scene. Like if you get super deep into like which like old MacBook is better than the next one, like ultimately it like I just need my tool to get out of the way and do what I need it to do. There is like I keep I think about this a lot 'cause I I work in an AI company. I I talk to a lot of people that work at other AI companies. I don't think we've ever sat down and gone like, Hey, who's the best at using AI? Like that's just not a thing. Even to this day, we still talk about who the best programmer is or who the Like the best programmer, like especially because I work in like a medical technology company. I'm gonna keep it vague. I'll tell you after the show. ⁓ it's fine. But I work in a medical technology company, and a lot of these companies, like, it's cooked, bro. Like to access their API, it's so tightly knit that you can only access it from a VM at a certain hour in UK time, for example. No Chat GPT, no Claude. You're gonna have to raw dog that, bro. Like, you're gonna have to like there's no Postman in that. You're gonna have to curl command on Whatever the Windows thing is, the Windows thing I think they call it like bash something, like bog bash or I don't know what they call it in Windows. Like they call their command palette something else.

speaker-1: Yeah, you got PowerShell. PowerShell, yeah. Yeah. Command.

speaker-0: How to write like curl commands in PowerShell and hit the right API endpoints. That is the outcome of being super good at software engineering and knowing a computer. But being like a top level AI user, like what, you took water away from like a average plumber family in Idaho. Like they can't drink their sink they can't drink from their sink anymore. Like now they have microplastics because they're just buying bottled water.

speaker-2: Ca can I c I have a hot take ⁓ on the situation. so the the people who say that learning like that AI is a skill, I just think that they have no skills. Because that's what it must look like for someone who's never had to learn a skill. Like, you know, you have a bunch of people who've just who gone to like business school and a lot of like not to then there's probably some good business schools out there, but a lot of degrees are just kinda bullshit. Like You you c you can go to university for three to four years and never and not actually learn anything, but you get a degree in whatever n degree like ⁓ you know, like like business or communications or whatever. ⁓ and I think that the people who look at prunting and they think it's a skill, they j they must have just never learnt any skills. That's why it appears that way to them, right? Like if if you've learned how to like I started my first programming language was like basic and Pascal. that I started with when I like seven or something. And when you learn how to program, like when you learn with Pascal or with C or something like that, Ruby, which is a very enjoyable language to program in, it's so trivial to learn. It's like you can learn it can learn in like a few hours because you already know how all the underlying stuff it's written in C. You already know how all the underlying stuff works. No memory management, just very basic ⁓ syntax, very charming. But very easy to learn because you have the underlying skills. But if you've never touched a computer in your life, Ruby must seem like a massively daunting task. ⁓ this is I what's like like I gotta like, there's no parentheses, like like you have to like learn all these all these words. It must feel like like almost impossible to do. And I just have the impression that a lot of the people making decisions in business situations just have just never done any real like studying in their life. They went to uni, they have a degree, they think they're smart, they think they have skills, they might have some social skills, which is how they got into these positions. When I learned that most of the AI bros are not programmers, like people in charge of AI companies are generally not programmers, that blew my mind. They're all just business bros. It's incredible.

speaker-0: Even even influencers, right? There's a there's this on there's this AI influencer. He's big now. Like he's huge now. I think he's almost like f half a million subs or something. The other day called Cash Cache.

speaker-1: I love that. ⁓ That rule's actual.

speaker-0: Like when I was hearing to this guy, I'm like, yo, this guy hasn't this guy hasn't even seen the face of a computer book, let alone like C S one ⁓ one. Yeah.

speaker-1: So interesting this idea of like here's the pr the the like the the thing I love to think about is okay, yeah, you prompted a thing to build an app or you submit it to the app store and you have a functional piece of software. Like I'm not denying that you can't build a functioning piece of software. I question its security and I question you know, there are things where I'm like, ⁓ i like the AI agent might have put like a comment that said, I'll do this later. Right. And you wouldn't know that because you're not reviewing the code. Okay, I know you can do that. I'm not denying that. But I don't think that like warrants respect. I don't think that you didn't learn a craft or a skill. Now, if I go and I prompt it and I say, make me a studio Ghibli picture of speak talking into a fork with a microphone, like I don't want I should get no respect for that. I shouldn't be I didn't do anything interesting. I didn't learn, I didn't practice, I didn't study, I didn't spend hours and hours of doing something. And it's not that not that that's inherently ⁓ what everyone in the world has to be doing necessarily to like but but like there's a reason we respect artists and musicians and people who do creative things because it's hard to do. And it's it results in something that is like shared humanity and enjoyable. If you just prompt something and it outputs a thing, whether it's software or a studio Ghibli speak, like that is not interesting. It is not enjoyable. It's not useful. And like maybe that influencer you're talking about, or just your general yes, we have the spoon out. Give me Studio Ghibli version, please. ⁓ AI do it. It's like what I it's all in service, I think, of people thinking it prints money, right? Like that's where it's kind of coming to is like either there's it's always coming back to money for those kinds of people who want to influence things, right? They either want to make money by telling people how to do a thing because maybe they're scared or nervous or excited, or they wanna like build a piece of software that will make them money and they don't know how to program and then they're pr prompting a thing. Or it's like for ego reasons, right? I think that. Well, for the business things, people people people want money. Understand. But for like ego things, like there's there's no skill. It's and I think that that's where it gets dangerous too. So it's like you don't know what you're playing with. And I think that's where it gets a little worrisome. And like it'd be like if I like made a car and I didn't actually understand how engines and combustion and all these other complex things work. And I said, Hey, big box, like drive my car, go drift, go initial D style, drift around the corner. Like you'd be like, No. I don't trust you to make a safe and secure car. Like that thing's gonna explode. And I think

speaker-0: That's that's like the future with three D printing. Like you can just print a car in your backyard.

speaker-2: But the difference in skill between because I think like there's a lot of really skilled engineers. And because I I know you mentioned, Brett, in your video, that you would be willing to just switch careers if ⁓ if you had to essentially use AI for for work. ⁓ I mean, obviously there are many jobs where you could use that being able to think algorithmically would help you in many, many different jobs. It wouldn't have to be software engineering, it could be adjacent to it. You still have the skills. But I I've always wondered what would happen if like if if like all the talented engineers became ⁓ tradies or like tradesmen, right? Like what would happen what would happen if like all the top and all the like the really good engineers just said, you know what, I'm gonna focus on being an electrician. I'm gonna be like a really good electrician. I think that would be a very like as the engineers get displaced from their position of like being high up to like being a tradey. What would happen to all the people already in that field who maybe got into it because they were like, ⁓ whatever, my dad was a plumber, I'm gonna be a plumber, so whatever. Like that I think that's an interesting point. Like what would happen to all those people who will in it will essentially be in a huge amount of competition with people with a lot of skills and the capacity to to gain skills very, very quickly?

speaker-1: Yeah, yeah. The in the abstract, like what software engineering to me is, is it's problem solving and it's like learning and studying how systems work and and then pulling in dependencies. When people are saying, ⁓ well, you just go to Stack Overflow. I mean, that's just basically what using AI is. That's actually not true. When you use Stack Overflow, you're actually seeing an example of something and you're learning and understanding it enough to apply it to your system or your need or whatever. Or if you build a library, right, you're making something reusable and easier. Yeah. I mean, I think that applies to so many different things. I think I think what would end up happening is if you had these people who are doing software engineering and you applied them to yeah, a trade or something, you would end up I think you would get a couple of different results. Or like if I I'll just speak for myself, not broadly, because I can't I couldn't say, but like you would either end up being like orchestrating a bunch of different like something larger and more complex, like maybe like a building is being built and it has or it has certain electrical needs or like I could imagine There is a ceiling and like a skill that applies there. And I think that you take those same ideas of problem solving and learning things and understanding it and like you could apply it to things other than software engineering. Like that's where I think those skills are important. And you lose it with just like, ⁓ I'm prompting. And I think also atrophies if you don't do it regularly. So even if you're like, ⁓ I only use AI for the boring stuff, like I think you're doing yourself a disservice. Or like if you take, I'll give an example as like someone's like, ⁓ you're you're now an an artisanal programmer and You know, everyone else's IKEA and their manufacturing stuff. Well, like IKEA stuff is terrible. It's like low quality, it's cheap, which is nice and that you can afford it. But like if you compare like an IKEA table to like a person who handcrafts a high quality table out of really nice wood, it's night and day, right? And I think that that I could imagine j going and doing something where you're like really focused on the quality of something. Like when people talk about the craft of software engineering and code that's easy to change and maintain and it's well architected, like Those concepts apply to things broadly and abstractly. And like that's why there's skills. It just so happens that we live in this world where still, but really up until recently, it was like this highly valuable skill where it was like, ⁓ you could have a Google software engineer pay them $100,000 and then they're generating Google a million dollars in revenue a year. It's like, you know, that's why that programmers are paid what they're paid, because it was making businesses a lot of money. But I think if you take those skills that software engineers have. Like real software engineers, you could apply it to so many different things and like it would be fine and and useful and and fun and interesting. So

speaker-0: Yeah, totally agree. No notes.

speaker-2: You mentioned Usagi, ⁓ I believe that's how it's pronounced, your MG. ⁓ do you find yourself working more or less on personal projects now that you've stopped using AI? ⁓

speaker-1: More. It's like that's what helped me realize, like, ⁓ yeah, I can do this still and it's fun and it's enjoyable and I'm learning and I I care about it more, right? Like that's what helped me reconnect with it. And I think that's something I would recommend to anyone who's feel feeling maybe a little disenchanted. Maybe you're not like full, like full AI, because that was something I got a lot of feedback on. It's like people are like, ⁓ I use it for just a little bit here and there or when I'm stuck or whatever. It's like, or I've gotten an email so it's like, I'm like so disenchanted with programming and like it's totally lost its fun. And I used to like maintain open source projects or contribute to them or and now it's just like ⁓ it's totally ruined it. I would say like reconnect with the craft. And you don't have to make it open source, because then you have to maintain it and it's like a lot of work. But like make some fun things, like program for fun and remember how that felt. And if you felt that way in the past, and it's a really it's a really beautiful thing and it's a really enjoyable thing. And that that to me is one of those things where it's like, well. If I am if I am gone, if I am cooked and go electrician path or whatever, like I'll still be coding stuff. Like this weekend

speaker-0: I was gonna say, I'll call home the code.

speaker-1: I I use this like tiny little MP3 player to listen to podcasts. So I copy, I drag the MP3 file over. And I need a way to like manage my podcast. So I like built a little web app that uses with my local file system to just like download a new episode and then I put it on my MP3 player. It's like that's the kind of stuff where it's like for me so much fun and reminds me of learning programming and the best parts of it. And like, sure, you could go and prompt something to do that and whatever, but like that's not fun. That's like the whole thing and I'm trying to like reconnect with and and share is like, ⁓ yeah, this is like a fun thing that's worth doing and it's a skill and it's enjoyable.

speaker-2: Is Usagi for the ⁓ playdate console?

speaker-1: So okay, so I wrote a book on how to program games for Playdate. I started this book right after my second kid was born, and this was before I started using AI. So I wrote this book and I released it in early access. And Playdate is this tiny little handheld console that's almost like the original Game Boy. Yeah, it's very well designed. ⁓ teenage engineering helped with the designing if you're a music geek and you their stuff.

speaker-2: Yes, yes, yes.

speaker-1: That you can make games for it with C or with Lua. And there's not a lot of written resources out there. So it's like, ⁓ I'll just like start taking notes and writing tutorials. ⁓ well, this would be a good book. So I wrote this book. And that got me in this like Lua headspace. And Lua is this awesome programming language for learning programming because it's really simple. You have one data structure basically, which is tables, and tables is like arrays, tables is key values. It's like this really wild language. And you have functions. And there's some other stuff you have too. And you have some basic types, but it's like Really simple, really powerful. Wrote the book, finished the book. I was like, does anyone read programming books still because of AI? But I was like, I gotta finish it because people are but paying for it already and stuff. So I finished the book. And then I remembered like back in the day in Rust, this was three years ago, maybe more, I made this prototype for Yusagi where it live reloaded your code as it was running. It just read it from the disk 60 times a second, which is actually fine because there's like a cache for your file system. ⁓ it's not like it's slow. And I was using a different language than Lua, but because I come off that. Playdate book, and Lua is, I think, really great for learning. I was like, okay, let me actually drop Lua into Rust. And there's a great package for it. It's called ⁓ Lua. Like I did nothing interesting for Usagi. There's like virtually no interesting engineering. There's like two things that are maybe interesting. But the rest of it is I'm like gluing stuff together and use Lua for that. And yeah, so I'm like, hi, I really love Lua. I think it's a un Unsung hero. I think people see that it's one-based index and they just write it off as like this is like nonsense. But ⁓ Lua Lua rules. Like Lua secretly rules. And I think that's like you know the meme where it's like the Jedi on one side and Jedi on the other and the crying Jedi on the top with this like huge bell curve. Lua is like you're like, ⁓ it's kind of interesting. And then you hate it, and then you're like, ⁓ wow, Lua is actually like galaxy brain, incredible. Especially when you like everything's global, which is horrible. Like it's not I'm not saying it's perfect. Lua is like foot gun central. Ruby's also foot gun central, though. So like, you know, you're covering foot guns with Ruby and Lua, but like You can do incredible things and keep it really simple and beautiful and ⁓ anyway, yeah. So It's made of lua. Love would be like if you were a person who wanted to like ⁓ pit everything against each other because that's like how our sicko brains work. Like Usagi is a competitor to love in a lot of ways, but love is much more complex. So I actually added this command in Usagi. It's called Lovify. So like if you make a game with Usagi, you can run Usagi Lovify and it output it converts it to a love project so that I don't have to add features that love has. Because love can do like anything. Like you can make commercial games like Palatro in So Usagi's like in the middle between Pico 8, which is this like little fantasy kind of like ⁓ entire game engine with like IDE and sound editor. Usagi's right in the middle. You bring your own tools, has a couple of like really janky tools I made. And then if you're like, ⁓ I want I want big league, I want to I wanna be on Steam or whatever, then you would go to love.

speaker-2: ⁓ also Lua Neovim by the way.

speaker-1: Yeah. I'm used off of Neil, but I don't this to go on for an hour, so I'll I'll shut up.

speaker-0: Actually, gonna say, I think a lot of NeoVim's success is the simplicity of Lua. Cause I don't know if y'all seen Lisp, but Lisp is scary, bro. Sometimes Lisp looks scary.

speaker-1: Yeah, then I won't open the can of worms 'cause Big Box has to leave, but I could talk editors for way too long and ⁓

speaker-0: Recurring guest for Brett? Surely we do it again.

speaker-1: Yeah, I'll just say this. I'll I'll leave a cliffhanger. I switched to Helix about three months ago and I

speaker-0: I have so many opinions about that. I love Helix. I have so many opinions. I love Helix. Like like it has replaced like, you know, when you're on like a server and you just like vie into something rather than like obviously an S3 bucket doesn't have NeoVim installed. I keep going like HX these days and I'm like, ⁓ shit, this S3 bucket doesn't have Helix installed. I'm like, I'm not gonna install Helix on some rogue S3 bucket in Romania for some reason.

speaker-2: Okay, well then ⁓ we we're gonna have to bring you back on at some point, ⁓ so we can get ⁓ deeper into editors. d before we go, do you have anything that you'd like to promote, anything you'd like to talk about?

speaker-1: Yeah, I'd like to promote just doing things by hand. Like thinking about humans. I think that's really important. I have a a YouTube channel. I'm Brett Codes there. I have and that's about like coding and the craft and and some horrible TypeScript videos. So if you're like in TypeScript web dev, a little bit of Ruby. I I have weirdly like these Ruby screencasts that were on the Odin project. So they have a lot of views and like people are like, I'm here from Odin and ⁓ that's cool. Not the language. That's a different thing. This was like software boot camp online. ⁓ yeah.

speaker-2: Yes, Odin, yeah.

speaker-0: Those are your videos, right? Taught me programming.

speaker-1: ⁓ Yeah. Yeah. So like that was that was I made those. That was a long time ago. But anyway, so that's Brett Codes. And then if you're interested in game dev, I have Brett makes games. And that's where I'm teaching game programming and ⁓ these days a lot more Lua. And I like a bunch of Godot videos. Like Well, you know, Godot's cool too. That's a game engine for people who don't know. But yeah, so you can check me out there. If you're interested in the Playdate book, that's like the only commercial thing I've made. It's at PD Book dot net and it's like it'll teach you Lua, it'll teach you game programming, it'll teach you ⁓ yeah, you make Sokoban, which is like a puzzle game, you make a JRPG like Final Fantasy, and yeah, I I like writing. I've written some free books too. ⁓ there's a free Usagi book if you go to book dot usagi engine dot com and I wrote a book on game programming with Ruby. So yeah, like I love I love teaching and sharing and ⁓ yeah, you can find me on ⁓ on those places.

speaker-2: Cool. Okay. We definitely have to bring you back on ⁓ to talk more about this stuff. But ⁓ Brett, thank you so much for joining us.

speaker-1: Thank you both so much. This was so much fun.