Hello, Welcome to React Round Up, the podcast where we keep you updated on all things React related. This show is sponsored by Raygun and produced by Top and Devs and Onvoid. Top and Doves is where we create top and devs so get top and pay and recognition while working on interesting problems and making meaningful community contributions. An Onvoid provides remote design and software development services on a performance basis, so clients only pay after the tasks are delivered and approved.
In today's episode, we will talk about artificial intelligence in the day to day life of a React developer. That is going to cover using AI tools to help you generate codes such as copilot, but also tools that are going to turn the designer files into a first version of the so that you can iterate later. My name is Lucas Paganini. Your host in a podcast. Joining me in today's episode is Chris Ruin, Hi everybody, and Peter Ossa.
Hi, E one. All right, let's get to it. So, Peter, Chris, I think my first the first thing that I like to understand is what row is artificial intelligence playing in the lives of both of you at the moment, Like, how much are you currently already using AI for, Which tasks are you not using AI for? And what things you haven't
used yet but you are definitely considered doing. It makes sense, So basically just a summary, what are you currently using AI for, what you are not but you are considering you just haven't had a time, and what you're just definitely not going to even try to use AI for. So maybe Peter, would you like to start Okay? Yeah, so yeah, yeah, is actually a very important, very useful to I as a developer, Like
I kind of reli on AI for most of my work. Now what it is actually very important to me is in the aspect of knowledge, right, So I'm a fan of actually, okay, using the EI to kind of get some ideas that I don't have. That's maybe some knowledge that I am true less about. So just for example, I see a strange method maybe yeah, well working with it like it could be and I see maybe it kind of a new method, and I'm like, oh, I don't know
what this is, and maybe I have no clue. Now, usually in the past we start going to yeah, obviously it doesn't admit you within the dogs, but you start going to start overflow to see its use starts using the Google and checking other places. Right. So what I do in then is something that does when usually charge you to be a cloud, right, it does give me like a kind of description like details, does a small
idea of the of of photose functions are or what those methods are. Right, So it's kind of helpful to just it's just like I'm assistant, right for me in the sense that it's it's for me like a knowledge bank. Right. So if I don't have an idea of things, I use that
now. One thing I don't use AI to do is kind of I don't let a kind of replace my thinking ability, right to because I think that's the aspect that a lot of people kind of hope with it to and be like, oh, you're working in you're working on you're working on a code base, and you'll be like, oh, generate this to do this and you just learn on the code on they are. Yeah, in the long grown, Yeah, you'll be efficient on just re both in the long grown.
If you're not really learning anything new, let's assure if or the challenge is something that requires you to learn something, right, you're just accidental?
Are you just generate it for you? And you just use so for self development, willing not the best like for especially for going developers as well, like if you're starting out with the menially bad at always here, if you're starting out because you need to learn, need to just learn the basics, and you need to actually learn and practice with the basics, right, So I usually I just more like an assistant too, to assist me in giving me insights about settings. I don't know. I also use it to kind
of an interactic sation or is there an application of this? Can I how can I use this in this kind of code base? Yeah, will you get the ideas, then you can always kind of modify the idea. So I think that's just how I use Yeah, I don't know if that answers your question. Yeah, it does make sense. I'm not sure if there
is a right or wrong answer. Is definitely how you're seeing it, So I think it I could really understand the applications that AI has in your day to day So I think, yeah, it made total sense to me. How about you, Chris? Yeah, so I think I'm quite similar to Peter so me personally, I have I do have a subscription to Copilot, happy to pay for that. It's been it's been working quite well. And I do use chat GPT sometimes, I guess, so those are the two
main tools I use. Just yesterday was experimenting with Google's Gemini, the new Gemini AI, so just just trying to see getting a feel for how it compares to Open AI stuff. And I guess so I'm I'm fairly agnostic. I don't know yet. I don't have enough experience to say if there's a tool I guess I wouldn't use. But for me, it's more it's more just a function of time and and workflow that I'm used to. So I
you know, I've been developing for a while. Now, I know my tools, and it's just a matter of saying, okay, you know does this you know, for example, you mentioned Lucas, there's there's tools that can generate perhaps designs just from Pigma files and stuff that could be very advantageous. But I just I don't have the time, I haven't explored it. So I guess yeah. More as to how I guess, I use them, and I think Peter pointed to a good point which I'm guilty of forgetting,
is that I think we're so used to at least recently. We're so used to thinking of AI as it just generates, It just produces stuff, stuff, stuff stuff. But what I forget a lot is that in a new code base you can throw in functions and classes and say like, hey, what's going on here and get a nice overview without you know, maybe pouring over five hundred thousand lines of code. So that's something that I think for for anyone using AI, that's something that you shouldn't forget. And yeah,
that's that's uh, I guess that's yeah. Again, I also don't know if I answer your question. That's that's kind of how I how I see it, okay. Uh. And actually, one thing that I am going to recommend for anyone who is already using Copilot or any other similar tool, is there are ways for you to kind of like when you're talking to chat GPT, you create contexts before you start asking the things you really want. So let's say that you are looking for another job and you want help
to review your resume. You're going to have some initial chat with the AI just explaining who you are, what do you want in your resume, to which positions you're most interested in applying to. So you're going to create this initial context before you start asking the AI to answer your questions, right, and with code we can also do that. And the way that we can do that is well, of course, your entire code base is already context
for the artificial intelligence. Like many many of those AI code writers are going to consider the context of your application, like is it a React application and Angular application? How is how are the other developers structuring the code? Those
tools are going to consider that. But one thing that you can do to give explicit instructions because some maybe you're worried about that, like I don't want the artificial intelligence to consider the rest of the cobase because the rest of the cobase is like very messy and I want to do something different, I want to reflector so it should not considered just that. So what you can do
is document your style guide. You know, you can just have a style guide dot mb file at route and be very very explicit about every single convention and best practice that should be used in that code base, even if they are not being used yet, and that is going to serve as material for the artificial intelligence that you're using to know what style of code to generate. So that's kind of like a quick hot tip for all of you using those
tools. Well maybe not quick, because you still got to write the style guide. Yeah, it's It's something that I think some teams don't worry too much about because they think, oh, like the developer, where we don't have developers joining this project, like there's no reason for us to stop and document the style guide. But there actually is, because that's also input material for artificial intelligence and also for every other developer that joins the project later on.
So yeah, I wonder if you guys have any other chips or tricks that you use to kind of fine tune the AI to make sure that it is going to produce code that you're actually going to use. Well, yeah, on my end, what I do. I don't know if I have like a peculiar technique too, but I usually try to engage in discussion. Like you said, start with the premise, will give this is, don't point this is what the problem is? This don't find be like an intellectual
way. I think we get to point whereby I kind of write some kind of commands, right, I think you also put it in the way for example, if I'm actually writing. Maybe I also just get an idea or something in the text, I put it in maybe human readable format. I think that would be the point. Or I want this to be human re double format so that it can be understood by someone with maybe minimum knowledge a
party subject. Same thing with code as well. Just okay, I don't you can I could just maybe put it to bad functions or use the fundamental like maybe just committees on the edge, do not use maybe things like some higher other functions like I'm using Java spets to like maybe if I have like a could stuff. So I think there's are also that I do as just belable from Marth those format it in the way there's kind of low level so that someone who is a beginner can understand and so on. Yeah, not
really much though, and also I don't know. I usually, especially for company projects, I guly tried to avoid putting company data into the equal because I think there was an issue when I don't know if it's from amazonaway there was a leak of setting company data due to I think I think it's from charge you TPOs, so I think there was a use about that. So I gently tize to avoid us putting company logic or data. I just put it out about logic just because of I could get like ideas. Yeah,
so what do you do as a christ? Yeah? I guess, I guess. One advice I would have is that you have to be specific. Right, So it's it's just like the whole you know, manager specification joke that we all know as as devs. If you just say, if you go to any of these generators and say, hey, make me a button, you could, I mean you can. Yeah, okay, you specify. Even if you specify React, make me a React component button, you
can get who knows five hundred different things? Right, So you it's it's best to you know, I typically specify, like you know, I want to React type script. You know these props. Uh, you know, be as specific as you can. And if you if you can't be specific, well then maybe, you know, maybe you need to even think about what what your feature is or what what you're building at the moment. Otherwise
you can get just as much junk out as as as you want. So yeah, yeah, that does make sense, And it's what you would expect interacting with a real person, right. You can't just give very little instructions and expect the other individual to really do exactly what you had in your mind, You're going to have to properly describe it. So yeah, I think that that does make a lot of sense. And quite frankly, I don't think most of the issues that are generally document it in a project are good
enough for an artificial intelligence to understand. I don't know about you, guys, but I've spent quite a few time like trying to understand what the creator of the issue actually wanted me to do. So if I were to just throw that to an artificial intelligence, I would be really surprised if it would if you were able to really understand what has to be done. So yeah,
yeah, that makes doutal sense. What about like things that you are considering to explore, Like, of course we have those generators right from figma to code, but is it something that you guys really think is ready to test and ready to try out? Because I got to be honest, like I'm a bit skeptical. You know, I haven't even really stopped to see thelity of the code that is produced by those generators, because every time I saw somebody making a video about the code that was generated, it was always
like terrible. So it was like cool at first, but then it generated some stuff that was like either completely different from what it was supposed to do or just like the code itself was just really not using modern best practices, you know. So because of those videos of other creators are very talking about this, I kind of don't even feel the need to test it out myself.
But I don't know, maybe I'm missing out on something really interesting, you know, maybe there is a tool that is pretty awesome at doing this figma to code transition already, and maybe I am just a dinosaur, you know, and I haven't tried it out yet, so I don't know. Are you guys also skeptic about this or do you think that there are indeed
tools that might be already good enough for this. And by the way, I'm really sorry about this because I really wanted to be here to listen to your answer for this, but I will have to watch you later on because I actually have a personal commitment, so I'm going to have to bail the show now. But but yeah, I'm I'm sure the audience is going to be fine, like they are in very good hands. But yeah, let me know what your thoughts are on this and I'll make sure to watch you
later. Chris Peter, thank you all right, Lucas take care. Yeah, yeah, I think I think I'm kind of in a similar boat. I am quite skeptical, but I think it may even come down to just language and itself, because you know, going back to this concept like make me a button, you can even be even if you do something more specific. I ran into this was one of my first experiences kind of you know, poking fun at at chat geep Chan said hey, it's wrong, it's
wrong. But I realized maybe that I was wrong or not specific enough. So I, for example, you can say, you know, give me, I guess we'll stay in jobscript, give me a typescript function that compares to a rays all right. So what it did was I think it just did Ah. It just checked every every element index by index. Right, So if the first one first off, they're the same link and then if each element in each position is the same. But is that does that mean
the arrays are the same in this case? Is it's like what the same or comparing arrays really mean? And and when you start going down that rabbit hole that starts to become a whole question of you know, for a design team, like what really it's It's like, what it is the language behind our design, our product that all this, I guess this contextual knowledge that's very hard to put into a text box, right you know what does Yeah, just even like just small words like what does this mean? The team?
Your team probably knows because you've been working on this this product or these components or whatever, but uh, you know, chat GBT or Gemini or whatever doesn't really know. And I think then then what comes into play is this whole context window. And Lucas was mentioning that that you can kind of
pre treat the output. But unless you're willing, you know, as an organization to more or less pace your whole code base there, which you know could be dangerous for some organizations or not desired, then yeah, then I think you'll have some trouble. And even then, I've I've tried to do those pre treating It does work sometimes, but once you get I mean I've I've been just using the three point five so it has a smaller context window.
But even then, once you've you've provided a lot of similar code, it can still mess up. Just because yeah, I mean it's coding is hard, I think even for an AI so yeah, yeah, yeah, that's autually dumble things. Actually, so for genuitos like for figmat to code designs. Yeah right, I think the one I'm kind of it is enthusagic about is the BWD out the one I think built out of IQ. I think the one that actually converts figmat So yeah, components cool. Yeah,
so it's something I actually want to try because I've done some demos. Yeah, because of it. And yeah, the code and the code generation is a bit better this compared to some other ones of scene as well. White so it's a bit better. Let me just say, and I think they are better. They are better explanations why they could It's actually that way, I think, right, So, so I think that's actually when I do want to but over our door, I won't really like for the other generators
of saying, yeah, the code could be messy. Best of all, they don't. They just they don' don't follow like big partices like oh, for example, like you said, you're trying to compare maybe a compare array or an object, and then it's kind of doing some kind of expensive corpulations, right, maybe so expensive, like and it may not be optimal for your could like it's not may not be optimal like for memory in especially we were working with Javasket, which were kind of I won't say it's limited,
but we work with browsers and with limited environments. Is splitly limited memories at the point, so it's not like I suld we have all the memories are so us kind of so right, So yeah, so that's for that as well. Then the code structure the way it's been with it, Yeah, I don't know if they just look kind of clank in my eyes. I really don't know why. It just looks too jumped off like right, So
if you it's something that, Yeah, the it's the code. The elements of the code kind of takes a lot of like I want to say Boz was of big functions and just piece it around out so qu too. Yeah, so far I think code quality wise. And then the output, yeah, I've seen some of them. The output could could be nice for setting designs whereby it try it gives you I think almost piece of perfect designs.
But there's this something I see as well that it's kind of off like think I also see generators like Setta in the ears, I try to kind of like build like a software, like a small sUAS product in in a few seconds, and you'll be like, oh, give it some discussion. I just needed to build this application for me, for example, build like a rider application and so on, and then it's kind of the output is kind of not nice. It's not something presented at all, like it's actually far
from the desired output to right. So yeah, I'm I'm really yeah. Most for the ones of so far generators, they are really not like I'm not, I won't say they are. They're actually improving as well. But then I'm most in anyone that is actually kind of so it's so fascinating, except mably, I do want to want to try which is boed as bidous generator. I think fiedmach cool. I think I just want to try that then give my opinion. Yeah, I think I'm in the in the same
boat. I do want to try these tools, but like I mentioned before, it's just a matter well, first of time, I need time to try them. But I think they're they're quite interesting. It's interesting you mentioned builder dot io because just yesterday I was reading Netlefi they make a yearly kind of review of tools that people have used and then people or i'm sorry, tools that they want to use in the future. In twenty twenty four and
builder dot io is actually one of the most desired tools to use. So yeah, it might be it might be quite interesting to take a look at that. I think it kind of one. One thing I was thinking of as as you were talking about about Builder, I owe and the code quality is not so good. As I was thinking about how organizations, how you there there was a there's a there was a podcast from a guy if you know him, they call him d h as David Heinemeier Hansen. Yeah,
it was with Tim Ferriss. And I always remember this quote. It's such a small, like little sentence, but I always remember this, and I heard this like five or ten years ago. He says, you know, when you build an application or even maybe even a design you make you basically said, software is you know, hundreds or thousands or tens of thousands, very small decisions in you know, different components, different style CSS files,
whatever, in back ends, and those are done over time. Right, you can't, like you mentioned, so someone wants maybe hey, just build me a sass app. But it comes back to language and these decisions. You can't just say, build me a sass app because there's there's ten thousand decisions behind this and the AI. Okay, the AI can maybe choose all those for you, but then you don't know, you don't know what those
decisions it made, and then you're stuck with this. Yeah, even if it could generate a giant app, you have no idea the code, all
the code behind it, or or what it looks like. So yeah, for me, it's it will be quite interesting to see if those tools like the the I guess they I'll call for now over promising tools will really work because this reminds me a lot of there was a lot of hype around no code tools, which I think they're valuable and useful, but it I think you always come to a point eventually where I think, hopefully for us, you'll always need a developer or an engineer to to look in detail at certain
small things and modify things like that. So but yeah, I guess one other thing, uh, it might be fun to talk about with with these AI tools is do you think do you think developers are almost getting lazy or or or kind of relying too much, too much on AI because I know I'll be guilty of that sometimes where you know, I have like a function idea or something, and I'll just wait for co pilot to give me some suggestions, right, and maybe I mean, maybe that's okay because usually I
think you mentioned it's doing something inefficient or not the best. But at the very least you have, you have some generated code, you you get some ideas. But what do you what are you feeling about that or do you think do you think developers are going to get too lazy? Yeah? Okay, Yeah, so I think so actually because in the future, like like I usually say, like in the future, in the future, dependimize on issues. Yeah, I love develop will be lazy to do that. Now.
That's why I usually recommend, like I usually tell like people starting out that don't know you see, as you're starting out, I think you're heah, you will actually much usefulul like developers who are actually learned to the basics who are kind of up there, like you're maybe you've learned so much, you've worked on so much projects and then you have an idea of setting things right at that point would just be like maybe supplemented for you an assistant,
right, like, yeah, the air could generate the function for it. But it doesn't mean that if you just if you actually think it up as well, you just that you just want something that is faster for you to work quiet. Maybe you just need something you just you just need speed, right too For kind of engineers that are kind of more maybe senior, intermediate and so on, rights, it's kind of good to be helpful for them
in the ficure to be more efficient. But for developers starting out, due to the way here is being kind of praised, I think the gend developers who are kind of getting interested into development recently may become lazy, like from these starts, because then, yeah, you have an issue, and and first of all, like I, I what actually like about stack overflow is the fact that it helps you to actually such and then look at all other
people's solution, right. It gives you perspective, right because yes, judge, if you're right, it could you could just generate one solution for you and that's it. And I feel like, oh, that is just the solution. But then you could go to stack overflow and then read comments and read codes and see different solutions and you will get an idea from any of them as well. So I think that that's what a lot of the groupers will lose ultimately, that ability to be to kind of that diverse idea mindset.
Right. So it's just it's in the future. It may just be like, oh, all our codes look similar because everyone is using Judge TP or CO pilots, Like I could just look so similar because you're just like, yeah, used co pilots for this, and there's no difference because yeah you could. I think you could look presently, you could look at some codes and then you'll be like, oh, this is the style of this person. This is the style of this person. This is how your likestor
abybe uses it with an imperative style or something else. Right. But then now when you're when a lot of people become more with land on AI, you these styles get blood, right, you really don't know which start before
I use it. So for the cop COOM developers really good to be good to make them lazy, right, because I know that disearch disearch for acknowledge, like getting some good mental knowledge is actually eveny important because then I usually usually I really use myself as an example, right, when oh, I just want to look for a function that does maybe something like if this sorting right, sort setting stuff, and then I'm trying to maybe do sorting on
my own, and then you guctually if I, oh, jamas kid had is sort methought, Wow, that's so nice. I could just use that, and you learn about this subt method and use that. Now initially you're trying to just maybe use if you have an idea of that, Oh there's a there's a method like this, and you just want to find bad boom way to sort it, and then you discover duty first we're searching that you
sw this answer a better. So that's what's usually what kind of going to start with our flow research does and a charge you to also those away book. Like I said, it limits the perspectives you use. So yeah, I think in the future or common future to make developers lazy kind of. But then it depends on the level of the developer, right, so if you're a senior developer, you just be more efficient, like so fast, because then you actually know what to do. We've had the experience you've had
you like the foundations. Yeah, before coming thevelopers they actually the actually miss out on those there. So I think that's just my opinion on that. Yeah, I think you've brought up a really good point that, Yeah, I guess like before any of these tools you would just go to stack overflow or somewhere similar, or maybe even before that, before stack overflow, maybe you would even have a textbook or something like that, right, Yeah,
or or at the worst case, somebody some program or somewhere. And I think that's a very important point for people starting out. And I recognize that in myself. You know, the meme or the joke is that you go you just copy paste the very first answer from stack overflow, right, which
of course is it could work. But as at least as I've grown as an engineer, you realize, like you said, you read the comments, you see people say, well we did this, and then we you know, we discovered that maybe wasn't the best way, and you know, all these contextual hints that you can learn just as much, if not more than, whatever the accepted answer is. Right, And the risk I think with
these AI tools is just input output, right. I mean, sometimes they will even say they'll give you like some hints saying like oh this you know this and this, but I'm not sure if it will. I'm not sure if it will be at the level where if you read like a full stack over stack over for post or even a blog post about something, where you can get all these insights and all this I guess experience, right, because the yeah, the job of the AI is just at least right now,
is just I give you the best output, right. I don't give you context, I don't give you experience. So I think this is really tough for beginners, especially because you also need you need to know how to ask the right questions. And if you're a beginner, you don't. You can't even you know, you can't really explain what you want, at least at least very clearly. So yeah, one other thing I did, I thought of what you're mentioning, where maybe in the future, you know, code
is looking similar. I I worry a bit because I in a in a Go code base. I've been using Copilot a lot, just because like and I think Go is it's very simple to write, but it can get very very large, and I have I have some some files are growing quite large, and I mean I'm doing a lot of computation, but I'm worried that if we always go down the path of just like yeah, just generate it.
It works. Sure, it works, that's great, But there may be I think I'm worried that in the future will miss out on kind of what software engineering or part of software engineering is, like you know these where you take a task and maybe not the first iteration, but eventually you get to a solution. That's how do I don't explain it, not really compact,
but perhaps like elegant, I should say. And I'm worried that if if, if you always have the AI in the background, just you know, spitting out what you need at the immediate moment, will just end up like with these giant projects and yeah, maybe they work, but but they're not I don't know, maybe they're less maintainable. That that might be something I'd be concerned about. Do you do you see any risks like that or do you have any any thoughts about that? Or maybe I'm just being too
too negative? No, no, really, Actually you're right a point on
that. Why to do you could get like, yeah, I usually look at some solutions from maybe charge you tpiano, and then I see the kind of boost that's what I think I mentioned that in the aspect that yeah, if you're trying to do something, it beats minimal on it kind of gives you so big and so large could down Like isn't there just like a better way of doing this and kind of yeah, if your experience actually maybe you've had the experience with that you're actually in there, Oh yes, maybe this
is what this is. I think this should be a better way that you maybe modify to yourself. Right, you can just rely on the maybe all the code just oh yeah, this just take this works right, I find probably critically could you actually do some modifications right, so because of your expertise on your level, right, but then if if like it's not controlled right,
the coulde bits will go very over boos. And then I said it's lost, like it's just like a lot of a lot of big codes, like a lot of like I said, was like it also is a lot of boss code, whereas like maybe it's just it's one weird as like a
metal that you've probably not used in the wire or something big. And then when everyone looks at it, it's not if it's not it's not easy to read, right, So I think that's actually a very big fear, like because like I said, it's it's then like the consequence of yea, everybody could looks the same at that point, which I think I've seen it in most could be that a lot could be that kind of work with AI like
that much. Could they just look simil like just try to compare. Oh I've seen this song, I've seen this, This looks you could even identify that, Oh maybe this is help from pilot or something or argy, this is something that it will generate, right, So yeah, I think you're at a point on that actually, and I'm actually here on point of that. So but then I trually have a question as well as we that in as for you, now, do you think I will actually replease developers?
Because I know that's actually that has actually been like a very very widepread notion like so when people went part of it, like I literally know a lot of people that will want to go into cooting fully because the few they could be replaced in the future. What's especiality like junior developers, So what do you think do you think I could actually replace us all? Like, what do you think? Yeah, I think I think it goes back to that
that DHH quote quote that you need to make a million decisions. I mean, you can argue this also for they said like lawyers and people are also going to lose their jobs. I don't know, because in reality now I'm talking like actually outside the code base, right, so your organization or your clients or whoever you work with, you know, whatever, these hundreds of emails, these these design decisions, or any part of the stack, the
network, the back end. There's so many small little things. I mean, for me, the only way that AI could replace a real human developer is if they were somehow there and conscious. I mean then you're talking about you know whatever terminator anyway, So unless you can completely replace a human, I don't I just don't see how. I mean, they even even if you have the best code generator ever, it can look exactly, it could even mimic people on your team where you say, oh, it looks like
that code was written by him or him. Even then it's still it's still only code. And I don't know, you can't replace I just don't think it can replace someone with experience or or yeah, you know, yeah, experience with multiple projects or languages and things like that. So that's that's currently how I am so bubbill stay maybe we have terminator in a few years. And yeah, I think I'll tell you I I agree with you. So
I think from what DH from what you have reference on DH. Right, it's actually right, I see could like software development as they're a lot of moving parts. Right, so for example, you could actually have an idea for a product, right, and then we were working at the product you actually find that, oh, this business model I was feeling it's kind of
sucks, right, kind of doesn't really work. It does work processing consumers, and so maybe I need to change that, right, those situations back and forth, you actually test and then sh whether your business model works. You actually gen le maybe like a formula or a concept around it. Right, those things are kind of I feel they're human driven, unless maybe there's an air like you said, that can do that, which, like I
said, is like a amult already. So if you just just take over the world, then don't even participate in my standard for something that kind of thing, right, So so just that kind of thing. So if they're like unless I are kind of like that kind of level, right, So yeah, it's I don't think it they will actually replace you ons, right, because then you do those situations back and forth concepts testing, because yeah, I know that most most products and most such products they were not actually
the concept of which they were created are like limeters. Like for example, maybe he had an idea, but it wasn't the idea that initially that came into life. There were a lot of changes and iterations, and then those changes and iterations also got reflected in coude as well, right, So you need someone to actually process that those changes and then put it in coude. I don't know if ai we could wish to that level for we right, changing business specification, I don't. I don't know yet. So in that
regard, I don't think it will replace people. Now for senior developers, obviously, the experiences, these experiences, they're like because you can't replace song with against the more because they have the main experience of the whole business of the old thing, right, So the main experience is also very very very important right in the business and the product, So it's something that they can't really replace, right. I think most signiors will be kind of even hired
more in their appiods. That's what I feel than than than ever, so they need most. Now, I think the aspect I'm having concerned with is with the genius. Now, I, for example, I just want to use an example of maybe things like like we used the website generator. Maybe like a ux or ux designer. Right, your job is just kind of
design pages, right, the flow and the experience. Now, a lot of people have argued, and I think some people had looked at a lot of people who had the reasons that you were able to kind of scrib that division. Now it's really sorts because so many the developers that come into maybe things like web development, they start with the fun thing, right, and we they start with the fund thing. They have to stablish history and the
design part. So this is something they have to start it. And then the kind of a lot of companies are kind of eliminating that starting point for people they can actually have the experience with, replacing them with air. So I think that was the aspect that was really really concerning for me. But then I actually then remember that, oh, yeah, there's the accessibility, right, because yeah, there's something about just creating the good websites and that
have been accessible for a lot of people. So I don't know if here it kind of works with accessibility or maybe degenerators and so on. I think that one may actually be human intervened, but how many like I don't know how many people actually companies or so kind of a concerned about accessibility as much as that right, and there's certain codes they don't check for accessibility and so
on. So yeah, so I think overall it will replaced developers kind of well, I think there's a rising concunt for me for people who are starting out, especially people are coming to each setting genio right now, now there's a point they were reaching the guenior that obviously I feel that he has come be pleasant because then at the end off they they are the ones actually implement
those business ideas. Right, A senior has an idea, it does't main knowledge one idea and then it gives it to a senior who actually understand it and then bring up something right. And also the gueniors who subsequently become the seniors of the future. So it's something that we need that what is going to happen. So yeah, but it's for beginners who are going to setting domains like I know some years have actually admitted some writers, some kind of
writers. So you see I've seen I think I've u seen the job for like setting on LinkedIn for AI, right, and or like what like does that like? Yeah? I think yeah, I tally replaced people to really like just some people. Right, So what do you actually think on maybe people like like who's or which part of web develimit do you feel that has been printed by maybe the sogio AI? Yeah, I think I think front end definitely, But even there, I I still think there's you mentioned like
these things like accessibility. Yeah, okay, even if companies don't care so much about it. For me, like when I build websites, I make it a part to meet those requirements, right, And so maybe for people out there listening you might not even know this, but there's even rules about color contrasts to meet certain accessibility rules. And you know, maybe AI follows
it, maybe it doesn't. But if I, at least for me right now, I would feel much more comfortable with a designer who knows that stuff like and has known it for years, right, has the experience of all that and and you know, color choices even I mean I've even seen I'm no design or expert. But I've even seen how you can look at how fonts are chosen, and usually they have certain reasons, like you know, even for uh, you know, muted text when it's a bit more gray,
Why why you use this? And all these these things. Unfortunately I don't have that experience, but I can't imagine in AI just isn't designed for that. Again, it's it's just trained on. Sure it might be trained on the best websites, but it doesn't have that that contextual knowledge. So but yeah, I guess I think because we're probably coming towards Dana, I think maybe to end on on like a positive note, I think the nice
thing. While it is hard to learn, I don't think you can learn domain knowledge from an AI, but I think there is hope for for the junior developers that it is a nice tool. For example, I start, I started practicing some rusts the other day. I've never written RUSS before, but I think it can help. It can really speed up just the initial like onboarding or getting a feel for the language. I mean, at the
very least, the ais are pretty good with syntax and stuff. I think just for the domain and and you know, the actual project experience you still need to do and probably I think probably forever you'll always need to do the old fashioned way where you just need to have experience, you know, as a junior and grow into that senior role. But maybe maybe it's a tool to help help them them speed up. What do you what do you think? What do you think about that? Am I over optimistic? No?
I think you guys one points as well. Right, So it's like I said, it's actually I initially stated like an assistant, right, actually fascinated. Right, So I think using the eyes as like a starting out,
like starting out to actually increase your speed and lending. But then like where I actually don't recommend you in it is when you are actually like maybe starting out like on the basics basics, right, like, oh, you're coming down from another discipline and you have no idea what COULDE then just want to get into air directly, It's won't really be advisable because you're coming into a new discipline and then you're trying to learn something new. You actually need to
do the work and research and find out what you need to learn. Yeah, But then for people like like for developers, like we have actually maybe worked on cettain projects and certain things that I think for us if I don't if you just want to learn something, I actually have the idea of maybe basic syntas of languages and how they are kind of similar, I could just use the air and like I said, just start learning something, maybe a
new language or a new framework. Quite but then from beginners, I won't really advise that because you need that knowledge. You need to get that knowledge about it, like that the main knowledge about that new language or learning or so. Yeah, but then yeah, there's hope actually for the junior girl. Like and I think I mentioned it as well, they are always going to be seniors of the future. And those seniors of futures are not going
to be the A definitely, they're going to be the senior developed. They're going to be the juniors a group of today. So obviously there's always going to be a chance for a junior to rise up to become the seniors. So it's something I think that I think that's that's like the biggest equipment for any one. Now they're actually trying to come in and they feel out their junior position. Yeah, so it's just for them to actually kind of work
work on themselves and try to learn a lot of things. Right, obviously you're going to the juniors of today, Bill seniors of tomorrow. It's just it's also principle, right obviously like that. So I think that's just what I feel about everything about it kind of Yeah, yeah, I agree. I think, Yeah, lots lots to come in the future will be exciting. Yeah, I guess I guess with that, Ah, should we move
on to picks or or whatever you want to support or mention? Okay, yeah, for picks right, I okay, for picks, right, I will actually recommend that build out of areos without of areas, yeah, generator right, So I think I'm trying to get the link figm out to react. Yeah. So it's something I actually want to kind of look into, and I definitely want a lot of people to actually look into it as well. I think I've kind of followed this CITU. I think it is miscal
who is like the the creative of Angola. Yeah, so I think without I think they're also behind quick I think could the jerascipt framework, right, So yeah, I think I will past the link in the chat that I think more about them, So I don't know I will. Yeah, I think it's it's naturally on fig y. Yeah, they have a fig moth too. It's only easy fig moth too actually, and I think they'll be linked to the always set on how to do it. So yeah, so
I think that's just my topic for today. And then yeah, so what what's yours. I'll also take the builder I owe as as a homework. But yeah, I just wanted to mention. I think it's finally time I might destroy my email inbox. But so, I have courses on my blog.
I call them full stack courses, and I've mentioned in previous episodes, so I try and focus on it's either like a very specific kind of niche thing but that's usually very helpful in multiple applications, or it's more of a kind of you know, not a full complete sas at, but but some sort of working functional app that you build. And I will mention for people who for any reason can't can't afford the course that I do offer them for more or less for free. You can even so the first few lessons or
up to almost half of the course, they're free on YouTube. This is all from my blog. I'll post the link. But then, yeah, if you can't get the full course for any reason, just just reach out to me. And that comes with an honor system. So you know, if you're a dev and you've got a job, I would like if you pay for my course and support you know, what I'm building so that I can make more of these in the future and help more developers. So, but if you can't just reach out to me, I'll get you the full
course one way or another. Yeah, that's that's all from my side. Yeah, that's also so I think there is I mean, not much content to this, so yeah, I think we are. We've come to the end of the podcast today, right, so all right, and I think maybe I could just leave some partying words for the genior developers, like just keep on keep on, like keep on learning, and don't be afraid of AI, because, like I said, yeah, it's artificial intelligence, but
they're actually like they're like the real intelligence, the human intelligence. They're the original no matter what, and you can't do this place