"Yes, We Use AI" —
But We're Not Talking About the Same Thing.
When I ask a business owner whether they've adopted AI, the answer is almost always an instant "oh, yes." For a while I took it at face value. Then I started to notice something: the standard I carry in my head for "adopted AI," and the yes they're handing me, are almost never the same thing. They're not lying — we're just measuring two completely different things with the same word.
The same word, opposite realities
They're not lying. The problem is that "adopted AI" has no agreed definition. A company whose staff occasionally draft an email in ChatGPT says "yes, we use AI" with the exact same confidence as one that has rebuilt its operations around it. Same three words, opposite realities — and the word is comfortable enough that everyone who says it feels they're in the same club.
So the "yes" tells me almost nothing. The real information is hiding one layer down, in what they actually mean.
The ladder hiding inside "yes"
There's a ladder inside that single word, and almost everyone who says it is standing on the bottom rung believing they're near the top:
1. Dabbling. Someone on the team uses ChatGPT now and then for an email or an idea. Most "yes" answers stop here.
2. Tools. A few AI subscriptions are paid for and used ad hoc — no system, no standard, nothing shared.
3. Workflow. AI is built into specific, repeatable processes — customer service, content, a part of operations.
4. System. AI is how the business runs: a single source of truth, standards anyone can follow, the work genuinely depends on it.
5. Leverage. AI has changed the economics — the headcount you need, the margin you make, the things you can now do at all.
When I ask the question, I'm picturing rungs 4 and 5. They're answering honestly from rungs 1 and 2. Both are true. They're just not the same ladder I'm standing on.
The one-question test
There's a cleaner way to find the real rung than asking "do you use AI," and it cuts through the comfortable yes in one move:
It's the same lesson an outage teaches you the hard way: you only learn what you truly run on when it disappears. If nothing of yours would break, the "yes" was about a tool you use, not a system you run on.
Why the easy "yes" is dangerous
The danger isn't being on rung 1. Everyone starts there — I did, every company does, and there's no shame in it. The danger is what the easy "yes" does to you: it creates false comfort. You answer "yes, we've adopted AI," you believe you've crossed the line, and you quietly stop — while the businesses actually on rungs 4 and 5 keep climbing and pull away.
That's the trap. The "yes" doesn't mark the start of your AI adoption; for most companies it marks the end of it — because once you believe you've arrived, you stop moving. Dabbling is a perfectly good place to start. Mistaking it for the finish line is how you fall behind while feeling perfectly safe.
The bottom line
"Do you use AI?" has become a useless question, because in 2026 everyone says yes and means something different. If you want to know where a business — including your own — actually stands, ask the questions the comfortable yes can't survive: What did AI change about how you operate? And would anything break if it disappeared tomorrow?
Published by IMA AI — June 2026. We run on AI at the system level, on purpose — which means we also know precisely what would break if it went away. That's the difference between using a tool and running on one.