There's a comfortable story going around boardrooms right now. It goes like this: We bought the licenses. We sent the announcement. We even ran a lunch-and-learn. But our people just won't adopt AI.
It's a convenient story because it puts the problem somewhere other than the corner office. It's also wrong.
When AI adoption stalls, the reflex is to blame the workforce: too set in their ways, too anxious, too slow. But look closely at what's actually happening inside organizations and a different picture emerges. Leaders fail their employees before employees fail their leaders. AI is not primarily a test of whether your people can change. It's a test of whether you can lead them through it.
The workforce is more ready than you think
Start with the data, because on this point the data is unusually clear.
McKinsey's 2025 Superagency in the Workplace report surveyed more than 3,600 employees and 238 C-suite leaders. Its conclusion: "the biggest barrier to scaling is not employees, who are ready to incorporate AI into their jobs, but leaders, who are not steering fast enough." C-suite leaders are more than twice as likely to name employee readiness as a barrier as they are to point at their own role. Meanwhile, employees were three times more likely than their leaders assumed to already be using generative AI for a meaningful share of their daily work.
The gap shows up elsewhere too. A 2025 Harvard Business Review analysis found that 76% of executives believed their employees were enthusiastic about adopting AI, while just 31% of individual contributors actually were. Gallup's tracking tells the same story: since 2023, frequent AI use among leaders has risen from 17% to 44%, while individual contributors climbed more slowly, from 9% to 23%.
Leaders are pulling ahead and then mistaking the distance for proof that their people won't follow.
The adoption gap is real, but it is not evidence of an unwilling workforce. It is evidence of leaders who are out in front, looking back, and misreading the people waiting for direction.
What looks like resistance is usually fear
When people don't move, leaders tend to read it as resistance. More often, it's fear. And fear is a different problem with a different fix.
The numbers on workplace AI anxiety are hard to wave away. Mercer's Global Talent Trends 2026 report, drawing on nearly 12,000 respondents worldwide, found that 40% of workers fear losing their job to AI, up from 28% just two years earlier. Pew Research found that 64% of U.S. adults expect AI to mean fewer jobs over the next two decades.
These are not people bored by AI. They are watching a technology reshape their field and wondering, quietly, whether they are being trained to automate themselves out of work.
Now pair that anxiety with what employers are actually providing. According to Jobs for the Future, only about a third of workers say their employer gives them the training, guidance, or opportunities they need to use AI, and that number has fallen by nearly ten percentage points. Maximum fear, minimum support. People are scared, they don't know what to do, and no one is walking alongside them. So they freeze.
There's a specific cruelty to this dynamic, one Harvard Business School professor Amy Edmondson has spent her career documenting. The people who most need to experiment with AI (those in routine cognitive roles) are often the ones who feel most threatened by it. You are asking the employees with the most to lose to be the most enthusiastic early adopters. Without visible cover from their leaders, most won't. Not because they can't, but because no one has made it safe to try.
The proof is hiding in plain sight
Here's what should end the "our people won't adopt AI" conversation for good.
In a global study by KPMG and the University of Melbourne, spanning nearly 48,000 people across 47 countries, 57% of employees admitted to presenting AI-generated work as their own without telling their manager. Roughly a third said they actively hide their AI use, often to avoid judgment or scrutiny.
Sit with what that means. This is not a workforce refusing to touch the tools. This is a workforce using the tools so eagerly that people are willing to do it in secret, taking on real risk with sensitive company data, rather than do it openly.
"Shadow AI" is not a story about disobedient employees. It's a story about people who found the value immediately and correctly sensed that their organization had given them no permission, no guidance, no air cover to use it honestly.
When your employees are smuggling productivity into the building because they're afraid of how leadership will react, the adoption problem was never theirs.
Leadership is the variable that actually moves the needle
If fear and the absence of permission are the barriers, leadership is the lever. The evidence for how much leverage a leader actually has is striking.
Gallup's research, drawn from roughly 100 million employee interviews, found that the manager accounts for 70% of the variance in team engagement. Not the strategy. Not the tooling. The manager. And the behavior that most distinguishes the best managers is coaching, not directing.
That same dynamic governs AI adoption specifically. McKinsey found that AI high performers are three times more likely than their peers to strongly agree that senior leaders demonstrate ownership of AI initiatives and personally role-model using the tools. The differentiator isn't the model you license. It's whether the people in charge are visibly in the work alongside their teams.
This is why the aggregate numbers on AI returns are so bleak. MIT's research found that 95% of corporate AI pilots delivered no measurable business impact. The failure mode, again and again, was not the technology. It was leaders treating AI as a product rollout rather than a change in how people work, over-investing in tools while under-investing in the humans expected to use them.
You cannot buy your way out of a leadership gap.
What leading through this actually looks like
No new platform required. A different posture. In our work at Thinking Backward, the same sequence separates organizations that move from those that stall.
1. Name the fear out loud, then remove it
You cannot coach a frozen person. The first move is to address what no one is saying: people are worried about their careers. Leaders who pretend the fear isn't there leave their teams to manage it alone, in the dark, which is exactly where fear does the most damage.
Say plainly how AI changes the work, what you're committing to in terms of your people, and what you expect from them. Certainty isn't the goal — honesty is. Naming the fear is what thaws it.
2. Make it safe to fail
Edmondson defines psychological safety as "felt permission for candor": the shared belief that you can try something and have it flop without it being held against you. It is the precondition for everything AI adoption demands.
And it is built one specific way: leaders going first. Share your own clumsy prompts. Show the experiment that didn't work. When the most senior person in the room demonstrates that not-knowing is safe, everyone else exhales and starts trying. When leaders hide their own learning curve, the silence travels downward.
3. Give people the outcome and get out of the way
Permission to fail without ownership produces dabbling, not progress. The other half of coaching is handing people the result and the authority to go get it. Gallup's research is specific here: the highest-performing relationships involve two-way conversations where goals are set collaboratively. Tell your team what success looks like, make clear the result is theirs to own, and trust them to go after it. People move when they're trusted to, not when they're managed into it.
4. Then give them the right way to learn
Notice that the tools come last. This is deliberate, and it's the whole philosophy behind Thinking Backward. Once the fear is gone and people are emboldened to act, the question is no longer which AI tool should we buy but what outcome are we trying to reach, and what's the shortest honest path to it.
Start from the result your team actually needs and work backward to the capability. Tool-first adoption is what produces the graveyard of unused licenses. Outcome-first learning is what makes the tools stick, because they're now attached to something people actually care about.
The tools are not the test. You are.
What AI has given every leader right now is not primarily a technology problem. It's an opportunity.
Your people are more ready than you think. They are also more frightened than you've let yourself notice, and many of them are already using these tools where you can't see. What they're missing isn't aptitude. It's a leader willing to learn alongside them, to make failure survivable, to hand them the outcome and trust them with it.
That's what separates the organizations seeing real returns from the ones still hunting for them. Not the software. The leadership.
AI empowerment is a leadership act. Your people are ready for you to lead.
