The average job has already seen 32% of its required skills shift in just three years, much of it due to AI adoption. And the salary bump is real, too. Just having one AI skill listed can boost a job's pay by 28%. That’s roughly $18,000 more a year.
But here’s something: eight of the top ten AI skills employers want are not technical.
They’re human.
Think communication, research, leadership, problem-solving, writing, and customer service. In other words: you need to understand AI, but you still have to be a person about it.
Why most AI training misses the mark
The rush to AI upskilling has been mostly technical. We see companies pushing training programs on prompt design or machine learning APIs.
And while that’s important for engineers, it misses the mark for the rest of us.
Here’s what I’ve noticed helping people with AI in real life:
- Most professionals don’t need to build AI—they need to collaborate with it.
- They get overwhelmed by tools because nobody taught them the mindset first.
- They try to "learn AI" the same way they learned Excel—by clicking buttons, not by applying it to real work.
A better way to learn AI: Build small, useful things
Instead of handing people a 10-hour course, what if we started by showing them how to build their own personal research assistant?
Or create a content engine that generates three drafts a week?
Or automate the boring parts of their inbox?
That’s the idea behind a little experiment I’m thinking of running: an AI Lab for Everyday Work. A place where we explore how to:
- Build personal board of advisors with ChatGPT
- Create idea-to-draft content systems
- Use image generators for presentations or marketing
- Turn a mess of bookmarks into structured research
Nothing fancy. Just real tools, used by real people, in ways that actually make work easier.