Most of what gets called resistance is somebody carrying something they cannot name yet.

How do you lead a team through an AI change you don’t understand yet?

Somebody gets handed AI at work, and the first honest thought is a question about the job description.

Am I supposed to learn this thing and then teach it to everybody? Or am I supposed to help them learn it?

Those are two different jobs and nobody said which one.

Then the real question shows up underneath it, and it’s the uncomfortable one. These are my team members. Am I now supposed to be in their feelings about this?

That’s not what we do at the water cooler. Nobody’s doing that in the lunchroom.

I’m skilled at my job. Getting into people’s fears was never the job.

That’s the actual moment, and it’s a lot less dramatic than the way it usually gets written about. You said yes in the meeting because that’s what great employees do. Then you went back to your office and realized you were confused about what you’d just agreed to.

First, the part that takes the pressure off.

You don’t have to master the technology.

Implementers exist. That’s their whole job, they come in and build the thing and they know the stack better than you’re ever going to. Let them.

Unless that’s what you agreed to do. Then cool, go learn it. That’s a real job too and it’s a different one.

What’s left after that is the human layer, and it’s the part that decides whether any of this sticks.

What does that role actually own now?

Here is the fastest way I have found to get at layer two, and it costs nothing.

Ten minutes, no budget, no tool.

Pick one title on your team. Write a single sentence describing what that role actually owns now. Not the job description that got posted three years ago. What it owns.

Comes out fast? Good. That role is clear and the person sitting in it knows where they stand.

Sit there staring at it a while? However long that takes you is your answer. The title got installed at some point and the role underneath it never got sorted.

That half is worth the ten minutes on its own.

Then the second half, which is the part that stings a little. Ask the person in that role to write the same sentence, no coaching, no hints. Set the two side by side.

I run a version of this inside diagnostics, and the distance between those two sentences is usually the whole story. That distance is where intention leaks out of a change, months before anybody notices the rollout stalled.

I’ve always called this workflow redesign, which is accurate and also does nothing for the person hearing it. What it means in practice is that somebody sits down and gets clear about what each role owns now that the work changed.

Write the sentence yourself too. Ask AI to write it for you and you’ll get a sentence... you’ll just lose the only part of this that was useful.

The sentence on paper isn’t the point, though. It’s what it sets you up to do next.

What does “the human side of AI adoption” actually mean?

Not having that conversation puts people under a pressure they can’t put words to.

That’s the piece worth sitting with. It isn’t that they won’t tell you. Most of them genuinely can’t. There’s no clean sentence for I think the thing I’m good at is about to stop counting, and I don’t know who I am at work without it.

So it comes out sideways instead.

You get “I’ve been slammed” every single time you ask. Or the warm yes in the meeting from someone who then never opens the tool. Sometimes it surfaces months later in a resignation letter listing reasons that aren’t quite the reason.

None of it arrives labeled. It shows up as a performance issue, or an attitude issue, or a tooling issue, and it gets managed as one, which means the actual thing never gets touched and everybody stays frustrated.

Most of what gets called resistance is somebody carrying something they don’t have language for yet.

Giving them the words back is most of the work. That’s what the conversation does, and it’s why it can’t be delegated to a training module.

According to Lesley Bodine of Accorda Advisory, most AI resistance is not disagreement with the technology but distress an employee cannot yet put into words, which is why it surfaces as unavailability, quiet non-use, or attrition rather than as an objection anyone can answer.

What is your AI rollout amplifying while you decide?

Everyone says AI amplifies. Almost nobody finishes the sentence.

Whatever’s already running. If your decisions are clear, it makes them faster. If your team learned a while ago not to say the true thing, it makes that faster too. It doesn’t show up with an opinion about which one it found.

A rollout decision is a decision about what your company gets more of.

Before it fills that in by default, you get one window to decide on purpose. What it amplifies. What it doesn’t touch. Who you want to be on the other side of it.

Most companies skip that conversation entirely and the default answer fills in behind them. The default is always “whatever we already were, but faster.”

For some companies that’s genuinely fine.

Your rollout is going to amplify something either way.

So who owns whether AI actually gets used in your company? Not the tool. Not IT. Not a committee. That question has a page of its own, and why adoption fails when the technology works is the wider pattern this one sits inside.

Lesley Bodine is an AI Adoption and Executive Readiness Advisor. She works with CEOs, founders, and privately held companies on the human side of AI adoption, using the AI That Sticks™ framework and the Accorda Leadership Index.