If you’ve worked in any large organisation over the past twenty years, you’ve watched this scene play out before.

A new way of working gets introduced.

The business case is solid.

The technology works.

Early adopters show impressive results.

Then resistance emerges. Not as outright rejection, but as something more subtle. Concerns about fairness. Questions about job descriptions. Requests for more time to evaluate.

The innovation stalls. Eventually it gets quietly shelved or limps along with partial adoption.

I’ve watched this pattern kill promising initiatives at Fortune 500 companies throughout my consulting career. Now I’m watching the same pattern emerge with AI adoption, but with one critical difference that changes everything.

The fairness trap

One of my recent LinkedIn exchanges on an article I wrote about the AI Valley of Death made me think about this more carefully. Colin Duff, an innovation expert I have known for many years, posted about the widening gap between AI super users and everyone else. His observation was sharp: most organisations haven’t made the “accountability leap” from treating AI as a nice productivity layer to treating it as a performance lever.

What struck me was how this connects to something I’ve seen repeatedly in innovation work.

When you ask someone to change how they work, they often respond with fairness language.

“It’s not fair to expect things beyond my agreed scope of work.”

“It’s not fair to change expectations without changing my job description.”

“It’s not fair to penalise me for not adopting something optional.”

This language is powerful because it sounds reasonable. Fairness is a value most organisations claim to uphold. But when fairness arguments are used to resist any change that creates discomfort, they become a shield protecting the status quo.

The European dimension

I work primarily with European companies, and there’s an additional layer here that matters.

Worker protections in Europe are strong. This is genuinely positive for many reasons. Employment stability, dignity at work, and protection from exploitation are values worth preserving. This is also what the EU AI Act aims to help.

But these protections also create structural barriers to rapid transformation. Even if AI enables someone to achieve 2-10x more productive throughput, their colleague can legitimately refuse to adopt it. The employment relationship makes it difficult to mandate new working methods, especially when those methods require learning new skills.

Some colleagues will embrace the change. Others will refuse to shift from a status quo that has worked for them so far.

This isn’t laziness or bad faith. It’s human psychology operating as expected. We are wired to prefer familiar competence over uncertain improvement, especially when the uncertain path requires effort and carries risk of failure.

Why this time is different

Here’s where the AI situation diverges from previous technology shifts.

When organisations adopted email, enterprise software, or cloud computing, the pace of change allowed for gradual adaptation. Resisters could hold out for years without obvious consequences. By the time adoption became mandatory, sufficient training and support had emerged.

AI doesn’t work that way.

The gap between someone using AI effectively and someone not using it creates visible performance differences within months, not years. One person produces detailed analysis in an hour that would take another person a week. One team iterates through ten versions of a strategy while another team debates their first draft.

These gaps are not subtle. They’re obvious to anyone paying attention.

Colin’s post identified where this leads: at some point, leaders will be forced to choose between protecting roles and protecting outcomes. That’s an uncomfortable conversation most organisations aren’t ready to have.

The same biases, accelerated consequences

The psychological patterns driving AI resistance are identical to those I saw blocking innovation for two decades. Loss aversion, identity protection, competence attachment, status quo bias. None of this is new.

What’s new is the timeline for consequences.

Previous technology resistance created gradual erosion of competitive position. Companies that were slow to adopt digital tools lost ground over years, giving them time to course-correct.

AI resistance creates rapid divergence. The organisations that figure out how to navigate the fairness trap and actually change how people work will pull ahead quickly. Those that let resistance win will watch their best performers either leave for more progressive environments or become frustrated operating below their capability.

What actually works

After twenty years of watching change initiatives succeed and fail, I’ve learned that forcing adoption rarely works and creates resentment even when it succeeds technically.

What does work is making the choice clear while providing genuine support for those who choose to adapt.

This means:

  • Creating visible examples of what’s possible when AI is used well, not abstract business cases but concrete demonstrations from people the resisters respect.
  • Providing substantial support for learning, not a single training session but ongoing access to help as people encounter obstacles.
  • Being honest about the trajectory without being punitive. “This is where we’re heading. Here’s how we’ll help you get there. Here’s what happens if you choose not to come along.”
  • Recognising that some people genuinely won’t make the transition, and planning for that reality rather than pretending universal adoption is achievable.

The uncomfortable truth

Colin’s post ended with an observation that “bringing everyone on the journey” made sense when change cycles were slower. With AI, some people won’t adapt fast enough.

He’s right. And that’s genuinely uncomfortable.

It’s uncomfortable because most of us want to believe that with enough support and time, everyone can adapt. It’s uncomfortable because acknowledging this reality means making difficult decisions about team composition. It’s uncomfortable because the people who resist often have valuable institutional knowledge and relationships.

But pretending this isn’t true doesn’t make it go away. It just delays the reckoning while the gap widens.

The organisations that navigate this well will be those that combine genuine support for adaptation with honest acknowledgment of what’s changing. Not cruel, but clear. Not punitive, but realistic.

The resistance pattern isn’t new. The consequences of that resistance arriving faster than ever before is what makes this moment different.