know your tools

Remember when everyone was obsessed with React, but half the developers couldn't write vanilla JavaScript? That weird era when you'd interview someone who could build a component library but couldn't explain how this worked?

AI is becoming exactly that. Again. And learning AI the right way matters more than ever.

The Framework Pattern Repeating Itself

A few years ago, front-end frameworks were the gold rush. People jumped on React, Vue, Angular — whatever was trendy. They learned the API, memorized the hooks, got good at the framework-specific stuff. But ask them to manipulate the DOM without a library? Crickets.

jQuery did this before React. Before that, other tools. The pattern is always the same: powerful abstraction arrives, everyone rushes to use it, most never bother learning what's underneath.

AI is doing this at scale. And it's worse.

Why AI Amplifies the Problem

With frameworks, the damage was contained to web development. You could be a framework tourist and still ship decent UIs.

The worst-case scenario? Your app is a little bloated, performance suffers, you can't debug weird edge cases.

AI doesn't have that luxury. It's not domain-specific. It touches everything — code, writing, design, strategy, decision-making. The surface area is massive. And because it feels magical, because you can get results without understanding the mechanics, people skip the fundamentals entirely.

Download the latest model. Learn prompt engineering. Ship something that looks impressive. Never question how it actually works.

You Don't Need a PhD, But You Need More Than Vibes

Look, I'm not saying you need to become Andrew Ng. You don't need to derive backpropagation from scratch or publish papers on transformer architecture.

But for fuck's sake, understand what you're using.

If you're writing code with AI assistance, understand how LLMs generate text probabilistically. Know why they hallucinate. Understand context windows, temperature settings, why certain prompts work better than others.

If you're building products with AI, understand the difference between fine-tuning and RAG. Know what embeddings are. Understand the cost model, the latency characteristics, the failure modes.

The Computer Analogy

Here's the thing that sounds obvious until you sit with it: if you write programs for computers, doesn't it make sense to understand how computers work? At least fundamentally?

You don't need to design CPUs. But you should know what memory is. What a process is. Why some operations are fast and others aren't.

AI is the same. If you're using it as a core tool — not just casually, but as a professional dependency — you owe it to yourself to dig deeper.

The Uncomfortable Truth

AI is powerful enough that you can get impressive results while understanding almost nothing. That's exactly why it's dangerous.

The gap between "looks good" and "actually works reliably" is bigger than ever, and you can't see it without understanding the fundamentals.

We've seen this movie before with frameworks. The people who learned JavaScript deeply could adapt to any framework. The people who only learned React struggled every time the ecosystem shifted.

AI will be the same. The people who understand the underlying principles will adapt as the technology evolves. Everyone else will be stuck re-learning surface-level tricks every six months.

Don't be that person.