Transcription
Is it still worth learning to program, or is it a waste of time because everything will soon be done by AI anyway? I can't remember ever being this uncertain about this topic. Therefore, in this video, we will answer precisely this question, even if you want to learn it not to work as a software developer. And by that, I mean I will try to answer it, because I am a software developer and not a fortune teller. However, I have worked in the industry long enough and know how to program, but I cannot predict 100% what will happen to software developers and other programming jobs in the future. Soon, I will show you how I am preparing for this AI wave, so that I can seize the opportunities and not be left behind. I have worked as a professional software developer for over 10 years, and even I haven't written a single line of code myself in three months. And therefore, I naturally ask myself the question, if I'm not even writing code anymore, even though I actually know how to do it, is it even worth learning it anymore? A year ago, I would have said that AI can in no way write complete projects, but today I am doing exactly that. What does this mean now for the role of software developers, if AI can take over writing code? You are probably watching this video because you don't know how to program at all yet, or have just started. But every professional software developer, and I'm counting everyone who earns money with it in there, will tell you that only a small part of the working time is spent writing new code. We will get to what the rest consists of shortly, and I will explain this with a very concrete example from my time as an app developer at Payback. It looks different for the classic job of a programmer. There, it's really just about translating the requirements into program code. So, exactly what AI is already doing for us now. The skeptics among you will say that AI-generated code is always complete garbage, and fundamentally, one can say that the code often works, but a professional would have written it differently. However, if you compare what the code looked like 3 years ago, which you copied from ChatGPT, and what is created by modern models today, even the biggest tin foil hat wearer will have to admit that this is a significant improvement. How this will develop in the coming years, one can really only speculate about, but the tools will never be as bad as they are today. Therefore, back to the job of a software developer. I worked as an app developer for almost 10 years at various companies, including Payback, which you have surely heard of, and there I most recently co-developed a large new feature. At Payback, there are actually two core functions. One is collecting points by scanning your digital card, and the Pay function, which I mainly worked on. There, you generate a QR code, scan it, your card is practically scanned, and payment is made simultaneously. At some point, a rather large project in collaboration with Aral was planned. The goal was that you could unlock a charging station with the Payback app, refuel, and drive off directly, meaning not going inside to pay. The billing would therefore be completely handled by the Payback app. A colleague and I were responsible for the implementation in the iPhone app. However, before we developed even a single line of code, there were countless meetings to understand the overall system and how we could integrate our app into this system. Because from the outside, it might seem relatively simple at first glance, but so many steps happen in the background, because the app first needs to know if you are currently at a gas station, at which gas station you are, which charging stations are available, then you have to authorize the amount in the app. It communicates with several systems, including the charging station, and so on. You can already see that a refueling process involves quite a few steps. My colleague and I therefore defined all the detailed requirements for the iPhone app during the planning phase. Only then did the second phase begin, i.e., the actual programming. So we really had a very long list of tasks that we worked through one by one as a pair. And although we knew exactly what to do and worked together very efficiently, this alone took about 5 months. This was also because we completely rewrote the entire component, but that's not so important now, but it took the entire development phase, just writing the code with testing and everything, 5 months. This sounds quite long, but all software developers know that major changes, including rewriting existing code that is a bit older, can occur. Logically, the payment function of the app is really critical, and it must function reliably and, above all, securely. Therefore, during these 5 months, we wrote many tests, which are also part of programming. After we finished, the third phase could start, which was after programming. This included deployment, monitoring, maintenance, and reporting. Because even if we tested everything extensively, errors still happen, which in this case would be found by several million active users. So it was up to us to find and fix the causes. If we look at these steps, it becomes clear that today, we could have significantly reduced the 5-month programming phase with the help of AI. However, the planning must be done correctly so that the AI can work properly. Otherwise, it only leads to the well-known AI slop. There are simply too many steps that non-developers cannot build alone with AI, especially when the project significantly increases in complexity. However, developers who know how to work correctly with AI benefit enormously from the tool. But what happens if the code in the production system is broken? User data is lost, the app crashes, or the system is paralyzed by a security vulnerability? Or even worse, data is stolen. Such things happen even without AI, but someone has to take responsibility, regardless of who wrote the code. And that's why it's important to understand code and be able to program. Now, not everyone wants to learn to program to work as a software developer. And especially for these people, it can be particularly worthwhile to learn programming, because you can't even imagine how many people still benefit from it at work. Let's take my own wife. She works in the marketing department of a large company and is part of a team of eight or nine people. And this team occasionally needs small programs or apps to make their work easier. So far, such tasks and needs are handled by developers. But it's often difficult to argue why a team of developers shouldn't take, say, two weeks to build something for, say, 10 people. In reality, such tasks and projects are simply not prioritized. The developers can work on something more important, and in this case, everything is always more important, and I know this because I've been on the other side myself and developed such software. But for internal tools, my wife and her colleagues can develop them themselves with AI, without understanding the code behind it. It's just for a small circle. If you can program yourself, it's of course significantly better, but often it's already enough today if you know how to use coding tools like Cloud Code correctly. Especially if the employer is open to it, you can make yourself significantly more valuable. What used to be proficient use of Excel or SQL is today AI tools like Cloud Code and Cursor. I've been doing YouTube for over 8 years, and people who have been following my channel for a longer time know that the only right way to learn programming is with your own project. Because you need to have the opportunity to apply what you've learned, and it doesn't matter at all whether you work with or without AI. Since AI will definitely become an integral part of software development, it is now essential to learn how to work effectively with this tool from the beginning. I no longer have the time to program all my apps by hand and have developed several apps that are used by many people every day. Both completely with AI, but also with AI support. This includes my Happy Tracker Streak, my Voice to Text app Whisper Bar, and most recently my To-Do app One Week. Each of these apps can be developed from idea to finished app within 30 days without any problems. In my App Accelerator, you will learn my complete system on how to develop really good results with AI. In six modules with over 50 lessons, I will show you all my steps. You will also receive my personal AI agents for Cloud Code, with which I publish apps that people really like to use in a fraction of the actual time. You will get my complete workflow in the App Accelerator via the link below in the description, and for a few days only, there is a significant discount. Ultimately, mastering programming skills and AI coding agents is something that will be not only useful but indispensable for everyone. Therefore, to answer the question, yes, it is worth learning to program right now.