What Even Are "AI Skills"?

Imagine this: It's performance review time. You've just gotten your review and on room for improvement your manager puts "Gain AI skills."

What does that even mean?

I keep seeing people ask this question. Or ask about some online school course on AI Basics. Meanwhile, people are losing jobs to "AI" and the ones left are being told to skill up on something nobody can define.

Here's my attempt to describe what AI skills actually means for anyone in a corporate role:

Understanding of AI basics - Think not believing everything it says. It's confident and wrong sometimes. Knowing that is step one.

Prompting - Think having it ask questions to help you guide. Not writing a novel of instructions, but getting it to pull the right information out of you.

Tool Knowledge - Awareness of what's out there for your use case. PowerPoint is a common want, for example. A lot of tools make PDF or image outputs, but you need specific integrations to get real .pptx files. Knowing what's possible and what isn't saves you from chasing dead ends.

Awareness - How others are using it, how old is your knowledge. I stay current by trying things I saw on Twitter 30-60 days ago. By the time I get to it, I know if it was hype or actually useful.

Applications - Being able to do X with it. For me, that's things like using Claude to automate variance analysis at work. Actual output, not just chatting with it.

I'm a little out of my depth on some of this, but here's my attempt for technical roles:

Architecture - Think designing agents to complete a task, designing guardrails, etc.

Models - Think picking between models/technologies.

Technique - Big problems, use X or use Y solution.

Development - Think using Cursor or Claude Code to help program solutions.

Research/Designing - Think fine tuning, reinforcement learning, modifying a model for a specific thing. I don't really know this stuff, other than a few wasted hours trying to fine tune things myself.

Domain - So, if you are in Marketing, doing all of the above, but specific to your area.

The takeaway: experimenting is how you gain AI skills, unless you want to go into research. Skip the course. Try things.

The point of having AI skills is still changing, but it may just be about becoming a future architect - someone who knows how to put the pieces together.