AI tools like ChatGPT are great for coding — they can speed you up a lot. But without fundamentals, you’re not building software… you’re just generating code. You still need to understand architecture, logic, and debugging so you can guide the AI, validate what it produces, and avoid ending up with a pile of code that “works” but isn’t maintainable, secure, or scalable.
Skipping the stairs does not mean you are above them. It means you have no idea what is holding you up.
LLM's have really levelled the playing field a lot. A lot of the annoyances of learning, and understanding, computer programming is now gone. You don't have to stay up at night, studying to understand the differences between for and foreach loops. All you have to do is be a demanding customer, have great motivation and a willingness to iterate, and you can achieve great things. Or possibly just spend hundreds of dollars on tokens without actually launching anything of importance. Time will show. But for those of us focused on building things. Those of us with tonnes of ideas and never enough time, LLM's have opened a floodgate.
I've gone up the steps multiple times one by one and still fall down them again anyway (but at least my arm hasn't merged with my leg..)
I think the AI generated image is saying something important silently. AI output will always have defects 🤝
Seems to be the new reality. Not sure of the fallout. Will the new developers end up learning the "old stack" or will the market end up accepting the reality of hard to find bugs being added at a faster rate? Or how about the vastly increased difficulty of adding new features or making modifications to the generated code? Maybe the whole software development paradigm will shift to the notion of disposable software. Rather than modifying your code to add or change features you simply chuck it and create a new one from scratch.
small list of what I used only html css and pure js for.. Built advanced systems in pure JavaScript, using the web platform itself as the runtime rather than relying on heavyweight frameworks. Created single-source orchestration logic capable of coordinating multiple heterogeneous systems from one language layer. Designed deterministic execution patterns in an ecosystem usually treated as dynamic and probabilistic. Proved that JavaScript can be used for serious systems architecture, not just websites and UI. Developed multi-agent / multi-process style coordination models with shared state mutation instead of traditional messaging layers. Explored concepts similar to distributed operating systems directly in browser/runtime environments. Built visual simulations of complex ideas (AI swarms, neural systems, sovereign OS concepts, infrastructure maps) entirely in JS. Turned browser technologies into a high-fidelity rendering engine for interfaces and 3D-like environments without requiring full game engines. Built prototypes that blend AI + orchestration + UI + distributed logic in one coherent stack. Used JavaScript as an innovation language, not merely an application language.
"What's the difference between my code and your code since we're both generating it?" Someone asked me this at work, and in my opinion, there's actually no difference. The quality, security, cost and stability of the result the code produces is the difference. I believe AI does have the potential, in the right hands, to build full-stack software with all of those qualities. And until AI takes that part over autonomously or companies start to believe it can, I think that's what employs me as an engineer still. Might be one foot out the door already.
Is that hand or leg?