Our 4WD barely gets driven anymore. It sits in the driveway, and most of the time that’s exactly where it stays.
But every so often I need it. And every so often that means edging the big white beast into a tight space in a busy carpark, on a busy day, often with someone waiting behind me.

I used to be able to do this. I’ve driven that car for over 10 years. I aways knew where its corners were without thinking about it, the way you do when you are so used to something… that unconscious competence.

But I don’t anymore. The feel for it has quietly gone, and what’s left is me in a very large vehicle and a space that is technically adequate, and the growing certainty that everybody in that carpark has stopped what they’re doing to watch.
I shuffle forward, then straighten. I shuffle back. Nobody is actually watching, but that makes no difference at all.

And here’s the loop: because it’s unpleasant, I avoid driving it. Because I avoid driving it, I stay out of practice. Which makes the next time worse.

This silly situation I put myself in came in mind yesterday when friend, and old colleague, was describing their team as resistant to AI.

What we’re actually asking people to do

The word “resistance” does a lot of quiet work in change management. It puts the problem inside the person. It has them as stuck and set in their ways. It assumes the person needs to shift their mindset.

But look at what we’re actually asking.

Someone has a process for writing a proposal. It’s been built over years and is no doubt automatic and efficient. The way they complete that process or task on autopilot can be key on a bad day, or week, to help them keep functioning. When the task has been compressed down to almost no cognitive cost they can do it while thinking about something else.

We’re asking them to take that, pull it apart, and rebuild it with an unfamiliar tool in the middle. Which means the task gets slower and more effortful. The task or process requires decisions again and the first few attempts will produce something worse than what they’d have made on their own in half the time.

That’s not a mindset problem. That’s an accurate cost-benefit read of a situation or part of their work.

The audience problem

Here’s the bit I think we consistently underestimate.

Being bad at something is unpleasant. We humans tend to avoid hard things, and doing things that we suck at.
Being bad at something at work, where competence is what you’re being paid for, is worse. And being bad at it where colleagues can see is worse again.

It hits hardest at the top of the room:

• The senior expert whose reputation is precisely that they know how to do this thing. Fumbling with a new tool in front of a junior they’ve been mentoring is a genuine status cost.
  
• The person who’s always been “good with computers.” Now the gap between them and everyone else has closed, and the identity they’d built is suddenly less useful.
  
• Anyone who’s been burned before. They tried a tool early, it was rubbish, they said so publicly, and now going back looks like admitting they were wrong.

None of these people will say any of this. They’ll say they haven’t had time.

And note that the audience is mostly imaginary. Nobody in that carpark cares how many goes I take. But the feeling of being watched shapes behaviour just as effectively as actually being watched, and it’s the feeling we’re designing around, not the reality.

And the part we talk around

Then there’s the one that sits under everything else.

If someone half-suspects they’re being asked to train their replacement, they are not going to experiment. Why would they? Enthusiastically automating the interesting parts of your role is not an obviously rational act, particularly if you’ve watched what happened to a team down the hall.

It won’t present as fear. Nobody says “I’m worried about my job” in a workshop. It comes out as “I just haven’t had a chance to look at it yet”, which is indistinguishable from apathy, gets logged as low engagement, and gets treated with more comms.

Which does nothing. Comms was never the problem.

I don’t have a tidy answer to this one. But I’m fairly confident the first move is being honest about what’s actually changing in people’s roles, rather than offering reassurance nobody can guarantee. People can tell the difference, and the reassurance costs you more credibility than the honesty would.

Time is a capability

One more, because it gets missed constantly.

You cannot learn a new way of working in a week with no gaps in it. Experimenting means a failed attempt, a second go, a bit of curiosity about what else it might do. All of that needs slack, and most people’s weeks have none.

So when adoption is flat, it’s worth asking whether anyone was ever given an hour in which adoption could physically occur. Quite often the honest answer is no. Perhaps you ran a session at lunchtime and then returned everyone to a full workload with a new expectation attached.

Adoption is a capacity question at least as much as a skills one. That makes it a leadership decision, not a training one.

So what actually helps?

If the barrier is habit and exposure rather than knowledge, the design changes.

Make it small.
Not “use AI in your role.” One task. Preferably a task the person actively dislikes, where the current version is so tedious that even a mediocre first attempt feels like a win.

Make it private.
The first attempts shouldn’t have an audience. Give people somewhere to be rubbish at it without it being a performance.

Make failure expected.
If you’ve told people this will feel awkward for a fortnight, awkwardness becomes evidence it’s working rather than evidence they can’t do it.

Have people go first who can afford to.
Not the most senior person in the room. Someone with enough standing to look silly and not mind.

Come back.
One session builds nothing, and the loop I described with the 4WD works in reverse too. One clumsy attempt, left alone, becomes a reason never to try again. The return visit is where the habit actually forms. And by then the tool has usually improved anyway, so there’s a genuine reason to have another go.

None of this is new, really. It’s just ordinary habit change, which we happen to know quite a lot about, applied to a technology we keep treating as though it should be self-evident.

I’ve started parking the 4WD down the empty end of the carpark. Longer walk, no queue, nobody waiting. Slightly ridiculous but this works for me.

Which is, I suspect, roughly what everyone needs. Not more information about how to park. Somewhere to practise without the audience. And don’t get me started on driving 2 different cars each with windscreen wipers with controls on opposite sides. Oh my goodness the times I either trigger wipers on a sunny day when meaning to indicate, or the opposite. Urgh. I should just stay here at my computer under a rock!