More and more AI-related articles and influencers are encouraging people to use AI, but we should pause and carefully reflect on what this really means.
I’ve been using AI to assist with development for quite some time. It helps me refactor code, perform code reviews, and write tests—greatly accelerating the development process. Sounds amazing, right?
Recently, I started using what’s been called “the most powerful AI on Earth”: Claude Code. It’s incredibly well-tuned for working with code and can integrate directly with VS Code or Cursor IDE to modify code in place. This allows me to focus on crafting effective prompts. When I provide the right prompts, the AI can complete tasks more efficiently and reduce the number of iterations needed. Everything seems perfect—on the surface.
Cherish your muscle memory — it’s the result of your experience and repeated practice.
As you increasingly rely on AI to assist in development, you’ll find yourself focusing more on how to give instructions — in other words, crafting the right prompts to get the AI to fulfill your needs with minimal revisions.
Over time, you may become a developer who only presses Enter, while your problem-solving abilities decline and your muscle memory fades.
It’s worth asking: if one day AI isn’t available, or you’re without internet access, would you still be able to develop on your own as an engineer?
Rethinking Development in the Age of AI
This article isn’t intended to discourage the use of AI in development, but rather to highlight that — alongside its convenience — there are potential side effects we shouldn’t ignore.
As the saying goes, “Too much of anything is just as bad as too little.”
Some of my developer friends have fully embraced AI, relying on it to the point where they simply define roles and rules, then leave the rest to the AI while they scroll through their phones or snack, waiting for the code to be generated.
Others take a more cautious approach, treating AI as an enhanced version of Google — mainly using it for research or assistance with code reviews.
There’s a stark contrast between these two approaches. While there’s no absolute right or wrong in how much you rely on AI, one thing remains essential: a strong technical foundation.
Because if you lack the ability to judge when AI is wrong, you could be heading straight into disaster.
Top comments (0)