AI adoption is an organizational project, not just an innovation initiative. Learn how leaders can build a culture of curiosity and safe experimentation.
Image by Evgeni Tcherkassk from Pixabay
In Part One of this series, I made the case that AI conversations often stay in the clouds. I compared this to the recent Artemis II mission, where the flight mattered but the bulk of the real work began when the vehicle touched back down to Earth’s gravity.
The bridge between strategy and implementation isn't a nice-to-have. It's the whole game. Coming back down to earth to the real, operational work of building that bridge is where things get interesting. Even when that bridge exists, a lot of organizations still stall. The strategy is sound and the implementation plan is reasonable. And yet, nothing much moves.
Almost always, the culprit is culture.
When organizations decide it's time to "do something with AI," there's a predictable pattern. Someone, usually a CTO or a Director of Innovation, gets handed the file. They're energetic, they're capable, and they genuinely want to make progress. So they start running pilots, experimenting with tools, building proofs of concept.
And then they hit a wall.
In my experience, AI adoption treated as only an innovation project is bound to plateau. When responsibility lives in one team or one function, it lacks the gravitational pull needed to fully realize its potential.
The people who control strategy, budget, and organizational direction haven't engaged or maybe haven’t been brought along. The people doing the day-to-day work don't feel ownership over the work or approach. And the leaders at the top, the ones whose conviction is required to sustain anything, may still be on the fence.
Real AI adoption is so much more than an innovation initiative. It's an organizational one. And that means it belongs to everyone.
Why resistance is rational
Before we talk about creating that gravitational pull, it's worth acknowledging something that often goes unsaid: most resistance to AI is completely rational.
People are asking reasonable questions. Will this change my role? Will I be expected to use tools I don't understand? If I experiment and make a mistake, will that be held against me? Who actually owns this, and are they going to help me or just evaluate me?
Leaders who acknowledge these questions honestly make far more progress. When people feel pressured, they become resistant; when they feel seen in their hesitation, they become curious. The organizations winning at AI adoption have made it safe to experiment.
Reps, not revelations
As someone who teaches barre and spent years as a competitive athlete, I know something about the gap between visible transformation and the work that creates it. People admire athletic performance but it’s easy to forget that the strength, precision, and apparent ease are built on thousands of repetitions nobody watched: small movements, done consistently, over time.
Culture change works the same way. There's rarely a single moment where everything shifts. There's no big a-ha moment that makes people genuinely comfortable with new processes or ways of working. What actually moves the needle is the accumulation of small signals: a leader who admits they don't have all the answers but is learning alongside the team; a failed experiment that gets discussed openly instead of buried; a staff member who tries something new and gets recognized for the attempt, even if it didn’t work out.
The reps are the point.
Governance that enables rather than inhibits
One of the most common ways organizations inadvertently suppress a culture of adoption is by front-loading governance.
The thinking makes sense on paper. Before we ask people to use AI, we should have policies in place. We should define the guardrails. We should be responsible.
The problem is that governance structures built before any real experience with AI tend to be built around imagined risks. They're cautious in ways that don't necessarily correspond with how the technology actually behaves in your specific context. When people encounter a wall of policy before they've had the chance to try anything, the message they receive is: this is dangerous, proceed with caution.
A more effective approach is to build governance concurrently with experimentation. Start thoughtfully, pay close attention to what you're learning, and develop your structures in real time as you understand the terrain. It sends a very different cultural signal: we trust you to be part of figuring this out.
What strategic leadership actually looks like here
In Part One, I talked about the necessity of driving AI strategy from the top of the house. That remains true. But there's a version of top-down leadership that helps, and a version that hinders.
The version that hinders: leadership declares AI a priority, assigns it to a team, and waits for results. The organization understands that AI matters in the abstract but has no sense of how it connects to their day to day work.
The version that helps: senior leaders visibly engage with the learning curve themselves. They ask questions in public. They share what they're finding useful and where they're still uncertain. They celebrate teams that try things, especially when those things don't work perfectly. They treat AI as something the organization is figuring out together, not something being imposed onto the organization.
This kind of leadership is hard. It requires genuine humility about not having all the answers. But it is the single most powerful message you can send if you want to encourage a curious culture.
Where to start
If you're a leader thinking about the culture side of AI adoption, here's what I'd suggest:
AI isn't just something your culture has to absorb; deployed thoughtfully, it can actively strengthen culture. When teams use AI to solve problems together, to surface ideas that might otherwise stay buried, or to remove the friction that quietly demoralizes, something shifts. Work feels less like maintenance and more like progress. That's a fundamentally different and far more energizing place for leadership.
After all, the goal was never just to implement AI. It was always to build a better organization.
