I’ve been thinking a lot about how people use AI tools, especially LLMs like ChatGPT. Something doesn’t sit right with me.
I use AI. Yes, I do.
People I interact with recently started asking me why I myself argue to not use AI but end up using it.
Call me rude, but it’s not my job to explain them individually. So this post is kinda a reply to them as well as a guide to others who might benefit from my mental model of using AI.
Let’s go with a very simple and yet almost day to day example. Someone types a vague prompt like “Write a blog post about TypeScript vs JavaScript”, hits enter, and publishes whatever comes back.
No big effort. It’s like skipping the gym and somehow gaining muscles.
That’s the steroid approach. Fast results, but no learning.
My idea is to use AI as a supplement. Something that supports the effort we’re already putting in. Like how protein powder helps after a tough workout. Or take Creatine as another example. You still have to lift the weights, do the reps and show up consistently. The supplement makes the process a little smoother.
So instead of just asking AI to do your job and replace yourself with AI, ask how you can learn from it. I’m doing this in almost any part of my life.
Take this blog as an example. I write the whole thing myself. It’ll be raw, unpolished and has flaws. A lot of them. I ask LLM to identify and fix the mistakes. It usually helps me fix grammatical and typo errors.
I won’t stop there. I ask the same LLM to explain to me “What went wrong?”, “How did you identify it?”, “How did you fix it?” and “How can I avoid it?”. In some scenarios, I’ll ask them to give me a list of resources where I can learn that specific thing.
That’s where the growth happens.
My initial blog posts were shit. Rephrasing it changes almost all lines. But now, my mistakes are limited to a few typos and basic grammatical errors.
This is the growth in my perspective.
If I used LLM to rephrase and post it back here, the blog post would’ve been just the same. But I wouldn’t have learned much.
It reminds me of the old days when we’d write code, submit PRs and learn from review comments. You’d realize, “Oh, I shouldn’t have nested that loop”. Over time, that feedback loop makes you better, faster and most importantly smarter.
LLMs can be just the same. A personal editor, a patient teacher and a third-eye. But only if we let them assist us, not replace us.
Before AI, learning how to write well or explain something clearly took years. You had to read tons of blog posts, copy what you liked, and slowly find your voice. But now, you can write something, feed it to an LLM, and immediately get feedback that would’ve taken a mentor or editor to spot.
That’s powerful.
But we lose all of that if we skip the effort. If we just hand over the work and copy-paste the results, we’re not building anything. We’re just outsourcing our thinking.
Learning doesn’t happen when you delegate the thinking.
So, I never say no to using AI. I use it daily. But use it like a tool in your belt. Not a magic wand.
It’ll help you do better, not just look better.
So next time you’re about to ask AI to write something for you, pause. Write it yourself. Even if it’s messy. Even if you hate how it sounds. Then ask for help making it better. You’ll be surprised how much you learn in the process.
And over time, you won’t just write better blog posts (or something you use AI for). You’ll become someone who can think, communicate, and create at a whole new level.
With no shortcuts.
Use AI like a supplement, not a steroid.
