AI Resistance Isn't a Training Problem. It's a Trust Problem.

The Myth

Last week, a Fortune 500 CHRO told me her company had spent $4 million on AI training in 2025. Adoption rates were still under 30%. Her conclusion: “We need better training.”

I hear some version of this every week. The myth is clean and logical: employees resist AI because they don’t know how to use it. Fix the skills gap, fix the adoption problem. Roll out a training program, hire an AI champion for each department, and watch adoption climb.

It’s the kind of thinking that makes perfect sense on a slide deck. It just doesn’t match what’s actually happening inside companies.

Why It Sounds Right

Training is tangible. It’s measurable. It gives you completion rates and certification numbers to present at the next leadership offsite. And there’s real evidence that skills matter: Anthropic published data this week showing that experienced AI users get significantly more value from tools like Claude, using them as “thought partners” for iteration and feedback rather than simple task automation. Peter McCrory from Anthropic called this evidence that AI is becoming a “skills-biased technology” that “might potentially reinforce differences in outcomes among those who have higher skills at getting value out of these tools.”

So yes, skills matter. But here’s the thing nobody wants to say out loud in the leadership meeting: most employees aren’t refusing to use AI because they can’t figure out the interface. They’re refusing because they don’t trust what happens after they do.

What the Evidence Says

ADP Research released its “Today at Work” 2026 report this week, surveying 39,000 workers across 36 countries. The headline finding stopped me mid-scroll: only 22% of workers worldwide feel their job is safe. Not “a little worried.” Not “cautiously optimistic.” Seventy-eight percent of the global workforce does not feel secure in their position.

And here’s where the myth falls apart. The ADP data reveals that the single strongest predictor of engagement isn’t AI literacy, training hours, or tool proficiency. It’s perceived investment. Workers who believe their company is genuinely investing in their development are 5.3 times more likely to feel secure. Among that group, 53% reported being fully engaged at work. Among those who don’t perceive that investment, engagement drops to 12%.

Nela Richardson, ADP’s chief economist, put it plainly: anxiety, not confidence, defines how most workers feel right now. And as her colleague Jay Caldwell noted, “AI is entering a workforce that is anxious… and the importance for HR professionals right now is not as much about the technology.”

The data gets even more counterintuitive. Daily AI users are four times more likely than non-users to report feeling less productive. The people using AI the most feel the least productive. That’s not a training gap. That’s a meaning gap. Workers are doing more but understanding less about where they fit in the picture that’s forming around them.

When the CEO of one of the most prominent AI companies says something like that publicly, and then an employee’s manager sends a calendar invite for mandatory AI training on Friday, you can understand the disconnect. The message landing in the workforce isn’t “we’re going to help you grow.” It’s “learn the thing that might replace you.”

The Reframe

The companies I’ve seen move the needle on AI adoption aren’t the ones with the best learning management platforms or the most training hours logged. They’re the ones where employees believe, based on actual evidence in their daily experience, that leadership sees them as people worth investing in. Not headcount to optimize.

That means the real question for a CEO isn’t “how do we upskill the workforce for AI?” It’s “have we given our people any reason to believe this transition will work out for them personally?”

The ADP data makes this concrete: when workers perceive genuine investment, engagement jumps from 12% to 53%. That’s not incremental improvement. That’s a fundamentally different company.

So What

If AI adoption numbers are stuck, stop buying more training seats. Start with a harder question: does the workforce trust that leadership is building a future that includes them? Because the data from this week, across 36 countries and every industry, says most workers do not feel that way right now. No amount of prompt engineering workshops will fix what is, at its core, a leadership credibility problem. The good news is that credibility, unlike technology infrastructure, is something every leader already knows how to build. It just requires choosing to.