The AI skills worth your time, and the ones that aren't

People ask me which AI skill to learn first, like there's one correct list. There isn't. But there is a real difference between a skill that still matters two years from now and one that quietly expires the day a tool changes its interface. Here's how I sort them, after 40 years of watching which skills held up and which didn't.

01Worth it: learning to explain yourself well

The people who get good answers out of AI aren't smarter, they're just clearer. They say who they are, what they need, and what "done" looks like, instead of typing three words and hoping. Whichever tool wins the next round of this race, someone who can explain a problem clearly still gets a better answer than someone who can't. That skill doesn't belong to Claude or ChatGPT. It belongs to you.

02Worth it: knowing what to double-check

This isn't distrust. It's the same habit you'd already use on a new hire's first draft: read it before you send it. AI gets confidently wrong sometimes, and the people who catch it aren't the ones who understand the technology, they're the ones who kept the habit of checking anything that actually matters.

03Worth it: doing one real task until it's boring

Not reading ten articles about AI. Running the same real task through it, the same one, until using it stops feeling like an event and starts feeling like reaching for a stapler. That's the actual milestone. Nobody gets there by watching more videos about it.

04Not worth obsessing over

Memorizing this month's list of clever tricks. Chasing every new model the day it launches. Trying to become the most technical person in the room. None of that is fake, and none of it is why people will come to you for help either. It expires fast, and there's always someone happy to specialize in keeping up with it so you don't have to.