If you’re new to software development and you want to learn AI, you might be better off learning how to use LLMs for app generation before you learn to code.
Vibe coding is your friend.
Do it until you feel how the systems respond. When your machine says it’s ready for production… ask it why it’s not.
You’re not making disposables like the traditionalists.
You’re playing an instrument. This is art.
Good or bad, your skill level is going to be a lot more visible than it was in the old days.
So build on that.
Experiment. Break things.
Turn the universe sideways, just because you can.
Notice what the system does by default. Top-load your brain with plan composition, decomposition, and breakdowns. Learn to understand what’s being said in your conversations with the machine, and what the words mean.
They might be shaped differently than you’re used to.
Your job is knowing what questions to ask.
Stick to languages that are easy to read, with verbose compilers or decades of clean documentation.
The worst mistake you can make is dragging in legacy abstractions and old-fashioned integration practices. The industry has been doing it wrong for years. If you want to see the wrong mindset in action, read the comments on most LinkedIn technical posts.
Some things still matter: good practices, separation of concerns, k8s & Docker, the basics of security.
We all learned to do this by doing.
You will too.
Plan, document, and track your work like your life depends on it… because it does.
Start with the fun parts.
Solve problems.
Tear them down.
Challenge your mind model.
Learn what “better” means and do it better next time.
Be cautious of advice rooted in “experience” outside of AI.
I believe in you.
You can do it.