If you are learning to code, you need to know this about AI

Published Jul 30, 2026, 5:00 PM EDT Ian is a journalist and creator from the UK living in Berlin, Germany. He's been into technology longer than he can remember, and has spent most of his career writing, creating videos, or interviewing people about the technology he loves. Ian is most likely to be found in the terminal, or running a long way for no discernible reason. It's a weird time to learn to program. Apparently, all code will be written by AI soon, but experienced developers aren't so sure. Even Linus Torvalds is not immune to backlash for calling AI a tool developers can use. However, beyond the argument about whether you should use AI at all, is how to use it to complement learning rather than hinder it. You won't learn to code by asking AI to do it for you, but it can be the perfect self-learning assistant. For some, learning is not linear, and information organization is as much of a problem as the learning task itself. Sure, AI can code now for a cost, but it can also help you learn and improve as a developer. I was, and remain, skeptical about AI adoption, but my experience with 'AI as a self-teaching aid' has been undeniably positive. It just requires restraint. AI can fix your documentation problem Exciting? No. Important? Yes. The first thing I do when trying to learn anything is to create my own documentation. It doesn't sound exciting, I know, but your own document highlighting each thing you learn in order is like a browser history for your brain. I was initially skeptical of AI use, but its potential to make sense of poorly organized notes and documentation was what first made it click for me. Whether you keep a document you update yourself, or ask AI to keep a document for you to download later, having your own reference is an incredible tool, and the basis behind the concept of having a second digital brain. The other benefit of having this document is that AI can use it to either quiz you directly on the subject, feed it into a purpose-designed app like NotebookLM, or create Anki decks for you, so you can combine different forms of learning and recall without the organization and groundwork that some find difficult. If you've ever felt like your general abilities are hindered by your organization skills, this is a perfect starting point. You aren't asking AI for the answer every time; you are building up your own progression tree of knowledge. Even setting aside the moral arguments about AI use, learning is a deeply individual thing. These examples are far from exhaustive and work well for those whose understanding is limited by attention and organization issues. Your mileage may vary! Taming mathematics The journey from terrifying squiggles to a lifelong skill Programming could be described as using mathematics to organize a bunch of stuff. Whatever you want to code, you need to learn how to use the math behind it. The wording there is important, though: using mathematics is different from fully understanding it or working everything out yourself. Computers make it easy to actually compute answers to math problems, but if you don't understand mathematical notation, how do you ask the right questions? This is where AI can help, both with the direct translation of notation and with interactive visualization of mathematical concepts. Not only can it help bridge the gap between notation and code, but most browser-based AI chat windows will run code directly to plot graphs and other visual aids. As an example, if you want to learn mathematical notation and how it applies to physical objects, a simple gravity simulator is a great learning project. Start with a program that simulates a ball falling under the influence of gravity. The process of taking mathematics, converting it to code, and then testing it either on the command line or with graphics is the perfect starting point. AI can help some beginners by gradually increasing the scope. Extending the program to simulate throwing the ball extends your understanding of acceleration and mass. Zoom out to the planet level, and you're talking about Newton and orbital dynamics. Where AI shines compared to traditional learning methods is the ability to ask questions at every stage to fill in the gaps. By asking questions about every element you don't understand, you'll build a fundamental skill from first principles. This is the key difference between simply asking AI to do something for you, and getting the help you need to learn to do it independently. Frameworks/Codebase walkthroughs Learn any framework A common stumbling block for all programmers is having to work within a framework, or within other codebases. For beginners, learning the language is key; anything else is unnecessary complexity. For experienced developers, the structure of frameworks, libraries, and already existing codebases is as good as their documentation. For each case, AI can help you focus on what's important, rather than everything possible. A perfect example of this would be a game engine like Godot or Unity. Having AI walk through the codebase of example projects or follow tutorials with you gives you a single point to ask questions and fill in the gaps in your knowledge. In the case of Godot, this is especially effective as each script, object, and scene is represented in easy-to-understand text outside the engine. In this instance, AI isn't replacing normal tutorials or references. It's just another set of eyes on what you are learning, ready to answer any questions you have. The same applies to many libraries, frameworks, and concepts. Whether you are learning React to create web front ends, Flutter for iOS and Android apps, or PortAudio for low-level C digital signal processing projects, having someone walk you through the setup puts you in a position to actually start making things. In many ways, this is the opposite of the approach in the mathematics example above. Sometimes it's best to learn the things you need to "know" to get building, and learn the pieces holding them together later. It can be the difference between giving up from being overwhelmed and learning the core concepts to build on. Learn the language of coding Get help understanding common programming patterns I had a common beginner's problem when I started coding. Different languages and styles all seemed very distinct. C was C, Python was Python, Unity C# had its own rules. It led to a lot of confusion. While there are many fantastic guides, things like object-oriented vs functional programming, composition vs. inheritance, and common patterns for interrupts and callbacks still felt very theoretical. For me, these concepts clicked long after my initial progress in writing code, leading to difficulty parsing other people's code and a lot of decision paralysis about how to write my own. A simple JavaScript program that showcases each coding style shows how powerful guided learning can be. Would it have been different if I could have asked an AI to guide me through identical scenarios using different code approaches? I don't know, but the interactive questioning element makes it easy to really home in on hard-to-understand (for me, at least) concepts. Again, this isn't AI coding for you. It's an aid to understanding how to teach yourself. The Fine Grain Effect Break down what's important, abstract what isn't Learning with AI isn't something I often see discussed seriously, and it's not hard to see why. Polarized voices dominate all discussion of AI use, leaving little space for anything else. None of the ideas in this article are groundbreaking, but they all lead to a simple point: AI really is a tool. Unless you decide to avoid it completely, you have to decide how to use it. I learn things in chaotic ways, and I avoided AI for a long time. Now, using it with discipline, I'm able to bridge the gap between my curiosity and limitations. I may still be skeptical of how AI is developing, but since I started using it, I've needed it less than ever.