Learning Without Shortcuts

LLMs and AI agents have made my life much easier compared to, say, ten years ago. Learning new topics is now faster, more accessible, and even customisable to the way I want to be taught.

But there’s a hidden cost.

Back then, learning something new meant dedicating real time and effort. I still remember wrestling with RegEx for months before it finally clicked. Or trying to grasp Git not just the porcelain commands, but the plumbing underneath. It took me weeks to build a mental model of how it all worked. That struggle shaped my understanding.

With today’s tools, you can simply ask an AI to explain these topics in minutes. Convenient, yes but if it had been that easy back then, I probably wouldn’t have pushed as hard, and the knowledge wouldn’t have stuck.

That’s why last week I tried something different: a week without AI. No LLMs, no coding assistants. Just me, my brain, and the old-school way of figuring things out.

The result? Surprisingly powerful. It was frustrating, yes but it reminded me of the “pre-LLM” era, when my first stop every morning wasn’t ChatGPT but Stack Overflow. Back then, it was my dopamine hit. Earning reputation points felt like a game, and trying to answer questions forced me to learn deeply.

I realised something important: “deliberate periods” without AI can help keep our brains sharp. Neuroplasticity grows through challenge, not shortcuts. Once in a while, it’s worth reminding ourselves that we’re not just prompt engineers we’re still software engineers.

Schedule a “week without LLMs.” Struggle a bit. Look things up the hard way. Wrestle with the docs. It won’t be easy, but that’s exactly the point.