I can write code faster today than I could a few years ago.
AI can explain an unfamiliar library.
It can generate a function.
It can suggest a f...
For further actions, you may consider blocking this person and/or reporting abuse
The line that hit me: "We can now become productive before we become good. And those are not the same thing."
I'm a beginner — I started learning Python about two weeks ago and I've been writing tutorials about it on Dev.to. And this is the exact tension I've been navigating, except I didn't have words for it until now.
When I started, I used AI for everything. I'd paste errors in, get fixes instantly, and move on. I felt productive. But after a few days, I realized I couldn't explain my own code. I'd fixed things without learning why they were broken.
So I changed my approach. Now I try to solve it myself first — even if I look stupid doing it. Even if it takes 20 minutes for something AI could do in 10 seconds. I read the error. I form a guess. I test it. Then, if I'm still stuck, I ask AI.
And the difference is real. When I ask AI after I've struggled, the answer actually lands. When I ask before, it just floats past me.
The "try first, ask AI second" rule is something I'm going to keep using. Because I don't want to become faster at being wrong either.
Great post.
This is exactly the kind of learning journey I was hoping the post would encourage. The fact that you noticed the difference between getting the answer and understanding the answer so early is valuable.
Taking those extra 20 minutes to form your own hypothesis before asking AI may feel slower today, but you're building something much more important: the ability to think through problems independently. Keep that habit, you'll be surprised how much faster you become without losing the learning along the way.
"Productive before we become good" is the best one-line summary of this I've read.
One habit that keeps the learning in without giving up the speed: predict before you paste. When you hit an error, write one sentence about what you think the cause is before asking the AI. Then compare. When you're right, you've confirmed a mental model. When you're wrong, the explanation actually sticks, because you had a wrong guess for it to correct.
The other one is for code the AI wrote: close the chat and explain the function out loud, or to a rubber duck. The parts you can't explain are exactly the parts you'll struggle to debug later.
Absolutely. “Predict before you paste” is a simple habit, but it changes the whole learning dynamic. I especially like the rubber-duck approach because if you can’t explain the code without the AI in front of you, you probably don’t understand it well enough to debug it independently.
The goal isn’t to avoid AI, it’s to make sure AI accelerates your understanding rather than replacing it.
the part you skip is reading the error yourself, and that's exactly what you need when prod breaks at 3am and the assistant has never seen your logs. what helps me is breaking a function it wrote on purpose and guessing what'll happen before i run it.
That’s a really good practice. Deliberately breaking the code and predicting the outcome forces you to build an internal model of how the code actually works, rather than just accepting the AI’s solution.
And your 3 a.m. production example captures the bigger point perfectly: AI may not be there when things go wrong, but your understanding has to be.
This is so true! AI makes coding faster but it's harder to actually learn the fundamentals when you're not struggling through problems yourself.
For anyone testing AI coding tools, I've been using jzstoken for my API keys. Makes it easy to test different models without signing up for each one separately. They give you $5 free to start.
Exactly. AI can remove a lot of the productive struggle that actually teaches you how to think through a problem. I think the key is using AI to accelerate learning rather than bypass it, try to understand the problem first, then use AI when it genuinely helps you move forward.
And thanks for sharing jzstoken. It sounds interesting for experimenting with different models.
As AI becomes better at answering questions, the skill of asking good questions becomes more valuable.
This article was generated by AI lol