I used to think being a good developer meant being good at writing code.
Now I’m not so sure.
AI can generate a function in seconds.
It can build a component.
It can explain an error.
It can refactor ugly code.
It can even build an entire small feature while you sit there drinking coffee.
So if writing code becomes cheap, what exactly are we getting paid for?
The uncomfortable answer
We’re not really paid to type code.
We’re paid to figure out what code should exist in the first place.
A client rarely comes to you saying:
“Please write 437 lines of TypeScript.”
They say:
“My users keep abandoning checkout.”
“My team wastes hours doing this manually.”
“Our website gets traffic but nobody books.”
“We need this system to talk to that system.”
Those aren't coding problems.
They’re problem-solving problems.
And that distinction matters more every day.
AI can build the wrong thing extremely fast
This is the part that gets ignored in the AI hype.
You can give an AI a perfectly written prompt and still end up with a beautifully implemented solution to the wrong problem.
The code can be clean.
The UI can look amazing.
The tests can pass.
And the whole thing can still be useless.
Because nobody stopped to ask:
Should we even build this?
That's where engineering judgment matters.
The skill that becomes more valuable
I think developers should spend less time obsessing over how quickly they can write code and more time learning how to:
Understand the business problem
Ask better questions
Break complicated problems down
Make architectural tradeoffs
Understand users
Debug systems instead of just syntax
Review AI generated code
Know when NOT to build something
AI makes execution faster.
That means your ability to make good decisions becomes even more important.
Don't compete with AI at typing
That's a terrible game to play.
If an AI can write a CRUD API in 30 seconds, don't spend your career trying to become the world's fastest CRUD API typist.
Use the tool.
Let it handle the repetitive stuff.
Then spend your energy on the parts where context, judgment and experience actually matter.
The future probably isn't:
Developers vs AI
It's:
Developers who use AI vs developers who don't.
And eventually, I think the real divide will be even simpler:
People who can tell AI what to build, and people who just ask it to code.
What do you think becomes the most valuable developer skill as AI gets better at coding?
Top comments (0)