Is Programming Knowledge Still Important?

Is programming knowledge still important? In one word, yes. But not in the same sense that it was 10 or even 5 years ago. In the age of agentic AI, vibe coding, and ease of automation, having hard programming skills seems unnecessary and a waste of time, so why is it still important? And no, I’m not going to say you need to understand the code that’s being written so you can catch errors and understand mistakes; because now, if one agent messes up, you can just have a different agent fix it, and that cycle continues. Although it doesn’t hurt to understand what’s going on there. In my eyes, I think being able to program and understand how these AI systems work behind the scenes allows you to utilize them to their maximum capacity and also understand their limitations. If you’re building an automation system to create content for your business, it’s easy to open up a model and prompt, “make me content for my Italian restaurant business.” Sure, it works, and you may have a mediocre flyer to post on your social media, but investing extra time to understand what’s going on inside the AI model’s “brain” can be a massive return on investment. So what does happen after you instruct an AI to “make a poster”? AI is trained on data, loads and loads of it, and that’s what it uses to provide you with an answer. It will look at things like: “what do Italian posters look like,” “what are common poster phrases,” and “what have I seen most in relation to this topic,” so you’ll definitely get an answer, just not your answer. Understanding that artificial intelligence, at its very base, is a prediction model. So entering just one layer deep into AI by understanding it’s a highly trained prediction model allows us to utilize it significantly more efficiently. Feeding it more context, examples of successful media, and what platform to post on already sets the stage for a better output. Ironically, in this way, it’s like humans: if I ask you what the probability is that it’s going to rain on the other side of the world versus asking the same question but telling you it has rained there for the past four days straight, I’d probably get a more calculated prediction from the latter. Going one layer deeper now; instead of just providing it with context through the prompt, you can build a system that automatically provides your model with your content page, analytics, audience demographics, etc. I know what you may be thinking: “well, I can get an AI to build that system for me too,” and you’re exactly right, but the programming knowledge you have allows you to come up with the idea to automate the data retrieval. It’s like basketball. Anyone can take a shot, but only someone who understands the game can design a system that consistently creates good shots. I’d like to wrap up this blog with a message by Peter Steinberger, creator of OpenClaw, when asked on Lex Fridman’s podcast whether he recommends that people should still learn to program, which ultimately inspired me to write this blog post. “You don’t have to know how to splice an array... but you can use your programming knowledge to move from one tech galaxy to another.”