AI has a huge impact on ... everything? Definitely also on how we learn things. And IF we learn things at all. And then there's the big question about WHAT to learn as a web dev in world where AI can already do so much!
Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.
Transcript
From the show. Tap a time to play from there.
What and how should you learn in 2026? And I think there are two main trends or two main things you should focus on. The first one is working with AI, so taking advantage of these AI And I know there are developers that just don't want to because you feel like the results you get out of using AI are bad code, you spend more time fixing
that code than, well, than you save by using AI, and that is absolutely fine. And just to be very clear, if you rely on AI too much, you will very likely get bad results. Your skills will disappear over time or will definitely not improve and the code you get will maybe require a lot of fixing. It's easy to end up in a cycle where you keep on prompting and prompting and you feel
productive and then you end up in a code base that's full of spaghetti code that's hard to manage, hard to reason, where you need to take a lot of time to understand it and then even more time to fix it. That is a real danger. I've talked about that before and exists. But I'm not saying that you should become a developer that does not write any code anymore and that just uses AI. I'm saying that you should become comfortable working with AI, leveraging AI assistants.
So you should try to learn to effectively work with Cursor, Claude Code, or whatever your favorite AI assistant is. Now, what do I mean by that? I mean that you should learn how to size your prompts right, so how to not ask for too much, how to provide the right context, to make a plan first before you even start prompting. You should learn what not to prompt for. You should learn which parts of the code you need to write or at least where you need to start write the code yourself before you
then maybe hand it off to a large language model and AI assistant. These are the skills you should develop can be harder than it maybe sounds. But the thing is, if you're not working on your own, in can, of course, ignore it, but if you're not doing that, if you wanna keep the job, chances are that your employer will want a workforce that feels confident working with AI. You will need this skill here if you wanna be
employable. Most companies will demand that. So that is something you should practice to be open-minded about that because it's simply more fun and you will likely get better results if you approach this open-minded. And again, it's not about handing off all the work to AI, it's just using it for some of the work, at least, and of getting a feeling for what you can use AI assistants for where you
maybe see, uh, some value you can get out of that. That, by the way, also kind of implies, I get, that you will do more code reviews. So that is kind of an implication here that you will need to get better at doing code reviews. You need to get better, decent, at analyzing the code that may have been generated by AI and about judging that code. And that is a skill. And I know writing code is more fun.
It's the same for me, but the role you have as a software developer is simply changing and I don't think it will go away. It won't change to just doing code reviews but doing some amount of code reviews analyzing the code that has been generated by AI, that will simply be important if you wanna be employable. That is my strong belief there. Now, of course, as I mentioned, if that's not a focus for you, you can totally ignore that or not, depending on what
you want to do. Now for the second point here. You need to become an expert generalist and of course I know that these terms kind of go against each other so let me explain. In a world where we have AI models that know a lot of things or that are able to produce a lot of code, where much of that code is only average, where they make mistakes, where they introduce
bugs or security issues in your code, in that world you wanna be an expert. You wanna be better than those models. You could say AI is here, I will just use AI for everything, but in that case you will lose your job because AI. If your skill level is the skill level of AI models, nobody needs you. Let me be very honest here clear. So you wanna be better than AI.
Now you can't be better than AI in everything. You can't be a SQL expert, a JavaScript expert, frontend and backend, a CSS expert, PHP. You can't be an expert in all these things so pick your areas and let's be honest. We did that before AI too. We became frontend developers, backend developers, uh, DevOps people. We had our focus areas before that too and that doesn't change and that's also not fixed by the way. It can evolve and change over time.
You're not locked in to never change your role but you should pick a certain area you find interesting and that can be whatever you want and then you should become an expert That is a strong recommendation. The good thing is you can learn with AI. Now I'll get back to how to learn a little bit later but AI can definitely be an accelerator including when it comes to becoming an expert because of course you can use it along the way to ask
follow-up questions and so on because AI has all that knowledge from all the training data, it has all that information from all those blog articles.It's just not always good at bringing that to the table when Vibe Coding, for example. But it can be a good assistant, but not the only source, and again, I'll get back to that when learning something. So you can leverage that. But what do I mean with the generalist part now? Well, because AI has that very broad knowledge, it's
easier as a developer now than it has been a couple of years ago to work on features or work with technologies you don't have a lot of experience with and I'll give you an example. I, for example, recently got more into doing Canvas stuff, web GPU stuff, animation stuff, and these have all been areas where I expertise or where I didn't have it. But I used AI to help me there. Now, I'm fully aware that the code I get from AI will not be awesome in all
cases, but it will be better than what I can So I can then use AI to get some code or to help me work on certain features and then I can dive in, do some extra research, read some documentation, and build up knowledge with help of AI and work on that code that was maybe generated by AI and improve it. So what I'm saying is I think it's a good strategy to be open to use AI to dive into technologies or features you
historically maybe wouldn't have touched because learning them would've taken too much time. Now with AI, that gets a bit easier and you can get a first draft, a first code draft from AI and then build up on that. And that's what I mean with generalist. It's easier to become somewhat capable at many things wi- because of AI. But again, you wanna use AI as a draft machine or as an initiator and then do your homework, dive a bit deeper,
and combine your expertise and your capabilities as a human to learn stuff with what the AI gave you. And that is how you can go really deep on core skills, a few core technologies, and become somewhat good at many more. And I think that will be a pretty good strategy also when we talk about employability, but not just when we talk about that. But instead also when we talk about opportunities that are here because of AI and I talked
about that in my other video. You, as a developer, can use your skills to build amazing stuff on top of AI. Applications like my Build My Graphic built to help me with generating graphics, infographics, other graphics, which I then also opened up to the public. Here, I combined my skills as a developer with to build something that really helps me know. And that is something you can do as an generalist, for example, even if we leave employability aside where this
also help. Now, what about learning with AI though? How will that look like in 2026? Well, you can use AI for everything. You can use just ChatGPT to learn, but that is not a good strategy. I would advise against that. Instead, just as with coding, use AI as an assistant for learning. Use it to dive deeper, to ask follow-up questions, but use it in conjunction with a course, with
documentation, with articles, or with books or whatever floats your boat. And I know I'm selling courses, so I have a strong interest here, but it is that combination of human-curated knowledge bases and learning resources where you get a clear guidance, where you get expert insights. It is that combination of that with the personalized touch AI can give you. You can go through a
course, for example, like my courses, and there will be parts where you maybe struggle. Other students struggle with other parts, so there will always be parts in a course harder to grasp for you. Guess what? You can go to AI, ChatGPT or whatever, and ask for further explanations or for examples or for maybe a coding exercise you can do. And by the way, I'm working on some pretty interesting stuff here to kind of bring that together so that you don't have to switch platforms.
But I'll share more once I have more to share. Uh, you can, of course, sign up to my newsletters below to be the but that is something where you can get the best of both worlds. And that is how I would learn and how I do learn. If I'm learning TanStack Start, for example, which I learned last year, I did that with their official docs, by reading some articles, by watching some YouTube videos, and then I used AI to ask follow-up questions. Especially also AI where I provided the
docs as extra sources, I just put that into the AI chat, to then chat about that and then get the best out of both worlds. And that is how I would learn w- and how I will learn in 2026, and how I would recommend to learn in And again, I'm working on more stuff here to make easier and I'm super excited to share more once I have more. But yeah, that is therefore, uh, what and how to learn in 2026.
And as always, let me know what you think of that. And again, I got more exciting stuff to share, forward to.
GPT 6 Astra was released. It's a good model, I have to say. Especially for 3D work it seems to be a real game changer. But due to all the hype, you may be disappointed...