Only 6% of Companies Train Workers for AI. Why?

Eighty-nine percent of business leaders say AI skills are critical. Six percent have started training anyone. That's the defining disconnect of 2026.

The numbers come from Metaintro's workforce analysis and match a pattern across multiple reports this month. DataCamp found 59% of enterprise leaders admit their organisations have an AI skills gap — while most claim to invest in training. The WEF estimates 59% of the global workforce needs reskilling by 2030. IMF managing director Kristalina Georgieva compared it to a "tsunami" hitting 40% of global jobs — 60% in advanced economies.

Everyone agrees the wave's coming. Almost nobody's teaching people to swim.

The $300 Billion Contradiction

Companies are spending $300 billion on AI systems in 2026. They're buying tools, building infrastructure, deploying models. They're not preparing the humans who have to use them.

A 2026 workforce analysis found 95% of AI pilot projects fail to scale. Not because the tech doesn't work — because organisations bolt AI onto existing workflows instead of redesigning jobs.

The typical corporate response? A webinar. Maybe an optional online course. A lunch-and-learn on "prompt engineering." That's not reskilling. It's checkbox compliance dressed as transformation.

The OECD's 2025 Skills Outlook showed the gap clearly: only 19% of adults without upper-secondary education take part in non-formal learning, vs 61% of university-educated workers. The people most at risk from AI are least likely to get training. The gap's widening.

Who Gets Trained — Who Gets Left Behind

The WEF's Future of Jobs report surveyed over 1,000 major employers. About 80% said they'd upskill workers. Two-thirds planned to hire for AI skills. Only 40% expected to cut headcount.

Sounds encouraging — until you check what "upskilling" means. The WEF's most in-demand skills are analytical thinking, creativity, collaboration, digital literacy. These take months or years to build. A four-hour "Introduction to AI" course doesn't get anyone there.

Fast Company went further, asking the question most plans avoid: can you retrain someone from a routine-task role into a high-judgment, AI-augmented one? Their answer was blunt. If filling "top 20%" jobs is hard with experienced candidates, it doesn't get easier by upskilling workers who've never done that work.

This isn't about intelligence. It's about the canyon between a two-day course and what these new roles actually demand.

The DoorDash Paradox

This week, DoorDash launched a "Tasks" app paying its 8 million couriers to film themselves doing household chores — loading dishwashers, folding laundry, opening doors — to train AI and robotics companies.

Gig workers teaching machines physical tasks. The same workers aren't learning new skills. They're generating data that trains the systems that may replace them.

DoorDash told Bloomberg the footage goes to "in-house AI models and those built by partners across retail, insurance, hospitality and technology sectors." Workers get paid per clip. The AI gets permanently smarter. The workers don't.

The reskilling gap in a single product. Investment flows to the machine. The human gets a per-task payment.

What This Means for Education

The pressure's hitting classrooms and universities. The IMF called for redesigning education so students gain skills that complement AI rather than compete with it. Online platforms report surging AI enrolments. Universities are testing AI tutors and adaptive learning.

But the OECD warns that without deliberate policy, benefits flow to the already well-educated. A university student can learn prompt engineering in a weekend. A factory worker whose job just got automated needs something more structured, more supported, and more expensive than what most companies or governments offer.

The WEF named the skills gap the biggest barrier to business transformation. Nearly 40% of job skills will change by 2030. Sixty-three percent of employers call it their top challenge. This won't fix itself.

The Real Question

The WEF projects a net gain of 78 million jobs by 2030 — 170 million created, 92 million displaced. That's possible. But it assumes displaced workers can move into new roles. Right now, 94% of companies aren't trying to make that happen.

AI hardware and software budgets are enormous. Budgets for teaching humans to work alongside it are, for most organisations, zero.

That 6% figure isn't a statistic. It's a policy failure playing out in real time.