These are my mental models of AI, still evolving. I hope a few are useful to you.
As I've developed my own AI fluency, my mental image of AI keeps changing — and with each new picture, I think about how I'd explain it to a colleague: both so they can consider it for their own work, and to save them a little time. These metaphors work for me. They're mental models, not objective truth — but I hope a few are useful to you.
It's just software — but unlike any we've had before. The twist: you don't read a manual, you ask it how to work with it.
It knows a tremendous amount about the world — but nothing about your world. Dividing the work between what it knows and what only you do.
It makes things — sites, code, decks, documents — but it's a maker, not a vending machine. Why the same brief gives different results each time.
A tribute to the thousands of individual people behind AI — and an honest look at why I still call it my "colleague."
A habit that runs underneath all four metaphors: I often ask the AI about itself — how it works, how its memory behaves, what's in its context window, which tools it can reach. These "meta" conversations are some of the most productive I have, and I hope you'll build the same habit. Pick a question below to see why I find it useful.