Companies are spending billions to train workers for AI. Most of it will fail

Walk into almost any boardroom this year, and the agenda is identical: How do we close the AI gap? Leaders are projected to spend billions to ensure their workforce doesn’t fall behind. They are confident. They are aggressive. And, in most cases, they’re failing.

According to Mercer’s 2026 Global Talent Trends study, 63% of executives view AI work redesign as their highest-return investment. But only 32% say their workforce is actually ready. The response to this disconnect is a predictable reflex: throw more money at training.

But training is a multiplier, not a foundation.

If you give an employee with the right foundational skills targeted instruction, the gap closes quickly. But if you give that same instruction to someone whose underlying behavioral profile doesn’t fit the redesigned role, you aren’t “upskilling.” You’re simply creating a person with an expensive certificate doing the exact same job they did yesterday.

The reason this cycle persists is that we are investing in the “how” without a clear picture of the “who.” Resumes are just history. Performance reviews reflect yesterday’s requirements. And AI “fluency surveys”? They mostly measure how a person feels about a tool they’ve likely only used three times.

None of these inputs tells a leader who has the adaptability or the learning orientation to thrive in a redesigned role. So, the training is rolled out indiscriminately, the results come back uneven and the C-suite is left wondering why the needle hasn’t moved.

The problem isn’t the training. It’s the sequence.

Consider a maintenance supply distributor I observed. They were under immense pressure to grow sales following a major acquisition. Rather than rushing into a massive training rollout for new account managers, they stopped to build a predictive model. They identified the specific behavioral patterns that separated their top performers from the rest.

When they eventually deployed the training, they didn’t do it blindly. They matched the coursework to the specific gaps they had measured. The results weren’t just “better,” they were transformative. That group drove 20% more in net sales, translating to an extra $89,000 in revenue per employee. That wasn’t a “new” investment. It was money already in the budget, being spent on the right people.

This is the step the majority of organizations are skipping. Mercer reports that nearly every executive (98%) is planning an org redesign in the next two years. Most of these will be drawn up in a vacuum, devoid of any real measurement of the people expected to execute them.

I’ve sat in rooms with HR leaders who are terrified of assessment because they fear it feels cold or clinical. But there is nothing colder than setting an employee up for failure by forcing them into a role that doesn’t align with their natural strengths. When we skip the diagnostic phase, we aren’t being people-first. We are being process-first. True empathy in the age of AI means knowing your people well enough to know exactly where they will shine, rather than asking them to guess.

The cost of this oversight isn’t just a line item on a P&L. It’s a trust killer. According to Mercer, workforce “thriving” is at a decade-long low. Employees are already anxious about AI. When we ask them to absorb redesigned work based on executive assumptions rather than evidence, we create a retention crisis before we see a productivity gain.

It is time to stop treating AI literacy as a spray-and-pray exercise.

Flip the script. Stop treating work redesign as a planning exercise that ends when the org chart is approved. Instead, measure first, redesign second, train third and communicate throughout.

And stop conflating technical skills with fit. A skills inventory tells you what people have done in a world that no longer exists. A soft skills assessment tells you who is built for the world that’s coming.

AI is forcing a pace of change that most companies are unprepared for. Spending heavily on training without first understanding the workforce being trained isn’t “moving fast.” It’s just expensive guesswork.

Before you sign off on your next round of role changes, ask: Have we actually measured the people we’re asking to change? If the answer is no, you’re just flipping a coin.

This article is published as part of the Foundry Expert Contributor Network.
Want to join?