The Skill Gap AI Was Supposed to Close (& Didn't)

Jeff Raikes, former Microsoft Business Division president and co-founder of the Raikes Foundation, argued in Fortune that the "talent debt" he warned about in April is now measurable. Microsoft's 2026 Work Trend Index, which surveyed 20,000 workers across ten countries, found only 16% of workers have developed enough judgment to move fluidly between directing AI and doing the work themselves, a group Microsoft calls "Frontier Professionals."

Key Points

• Microsoft's 2026 Work Trend Index found the most valuable skill in an AI workplace is critical judgment of AI output, not technical fluency, and only 16% of surveyed workers have developed it
• Gartner predicts that through 2026, eroding critical-thinking skills will push half of all global organizations to require "AI-free" skills assessments
• A RAND Corporation study published this spring found most students using AI for homework worry it's hurting their ability to think independently
• Community colleges enroll roughly 40% of U.S. undergraduates; add HBCUs and regional state universities, and these institutions educate most of the future American workforce, yet remain outside the national AI-readiness conversation
• A Pew Research Center survey cited in the piece found 70% of Americans believe higher education is headed in the wrong direction

The Gap Isn't About Who Uses AI

Raikes points to Morgan Stanley research showing that in industries with heavy AI exposure, output per worker rose while employment held steady, meaning workers were augmented rather than replaced. But he flags what that data doesn't show: those productivity gains concentrate among people who know how to direct AI, not simply operate it. That's a narrower group than "AI users," and it's the group the 16% Frontier Professional figure describes.

Two Different Fixes

Raikes frames higher education as pulled between two responses. One is an efficiency model: cheaper credentials, faster output, targeted job training, a reasonable instinct given that 70% of Americans already think higher ed is on the wrong track. The other, drawn from higher-education innovator Paul LeBlanc's forthcoming book "Reclaiming Purpose: The University in an AI World," argues institutions should use AI to free up time for the things hardest to automate: judgment, relationship-building, and reasoning through hard questions. Raikes doesn't treat these as competing visions. He argues a system that only optimizes for one will eventually fail at both.

Competency Training Is Winning Because It's Easier to Fund

Community colleges are launching applied AI programs with local employers, and HBCUs are building pipelines with Google, Nvidia, and IBM. Raikes credits the momentum but flags what most of it is actually building: AI competency, prompting, summarizing, running tools, rather than the harder, less measurable work of building judgment. Competency programs produce a clean credential. Judgment doesn't, so it loses funding priority even though it's the skill Microsoft's own data says matters more.

What This Means for Employers

Raikes's argument lands squarely on business leaders: the companies that come out ahead won't be the ones deploying AI fastest, but the ones investing in people who can direct it, catch its mistakes, and own what it produces. That's a workforce-development problem as much as a technology one, and it's worth building into how a company structures onboarding, training, and its broader growth strategy around AI adoption, not treating AI literacy alone as sufficient preparation.

For marketing and growth teams specifically, the same principle applies at smaller scale: a team fluent in AI tools without the judgment to evaluate their output is optimizing for speed at the expense of quality. If Raikes's framing holds, that's a gap worth addressing directly. Our AI marketing services work is built around exactly that distinction, closing the gap between using AI and knowing when to override it.