Don’t Skip the Hard Part

These days, when I sit down to design a system or build a feature, I catch myself opening an LLM first. I type a rough prompt, and boom: ready-made answers, architecture, even code.

And honestly? It feels like cheating. It skips the part where real understanding gets built.

LLMs are great at output, but terrible at teaching depth. You might get working code, but you won’t know why it works, and the database designs that come with it fall apart the moment you ask about trade-offs or performance under load. And when things break, you won’t have the intuition to debug them.

You can’tunderstand history by watching a movie. Sure, it tells a story, but it’s shaped by the director’s bias and stripped of complexity. If you want to truly understand history, you have to read the books. You have to explore the causes, the politics, the context, not just the final events.
That’s when it starts making sense.

The same goes for tech. You can’t learn system design just by reading LLM answers. You have to wrestle with real decisions:

• Should I use Redis here or not?
• What breaks if traffic doubles?
• Why does this query slow down after 10K records?

Struggling through those decisions is what sharpens you. That’s what builds your engineering brain.

I’m not saying don’t use AI. Use it. But don’t let it become your brain. Use it like a whiteboard partner, not a guru. Because real developers, real architects, they’re forged in the mess. They get there by working through the why.