Most companies are trying to close their AI skills gap the same way: hire more specialists.

There’s a problem with that plan. According to a new OECD report, “Skills in the AI age,” workers with advanced AI skills make up only about 1% of the overall workforce. Every company chasing that same 1% is competing for a pool that was never going to be big enough to go around.

The Talent War Nobody Can Win

Business AI adoption nearly tripled between 2021 and 2025, from 7% to 20% across OECD countries. About a quarter of workers were already exposed to generative AI in their day-to-day roles by 2024. Demand moved fast. Supply of specialists did not.

Half of the companies that haven’t adopted generative AI point to the same reason: they lack the internal skills to use it. And when governments look at where AI training money is actually going, the split is backwards. Of the OECD countries funding AI-specific training, more of that money went into building AI professionals than into building AI literacy for the general public.

The OECD’s own framing is direct about what most workers actually need:

“Most workers will only need literacy to help them use and interact with AI systems effectively.”

Not some. Most.

Why the 99% Matter More Than the 1%

It helps to think of workforce AI capability in three tiers.

• Confident everyday users: people who use AI tools safely and effectively inside the job they already have, with minimal need to understand how the model works underneath.
• Internal champions: the “translators” who turn that everyday use into real workflow change, closing the gap between having access to AI and actually working differently because of it.
• Specialists: the small group who design and maintain the underlying systems. This is the same 1% this article opened with, valuable, just never big enough to be the whole plan.

Most organizations only need to develop the first two tiers well. The specialist tier stays small by definition, and trying to hire your way into it is a losing bet.

This isn’t just a hunch. A joint report from Udemy and Indeed, covering hiring and upskilling data from 2023 to 2025, found that just 4% of job listings mention AI, while AI already accounts for 67.5% of employee upskilling effort. Employees are already trying to build AI capability faster than companies are hiring for it. That points to a training and workflow gap showing up disguised as a hiring problem.

That gap shows up in confidence too. A 2025 study of chief learning and talent officers by i4cp found that 79% of organizations say their workforce is unprepared, or at best only somewhat prepared, to use generative AI at work. Yet only 18% of those organizations actually offer generative AI training to most or all of their employees.

The demand for capability is there. The supply of structured training is not.

An Old Lesson From Tuskegee

Booker T. Washington was born into slavery. By the time he founded Tuskegee Institute, most people had already decided what a formerly enslaved population could do with their hands: basic labor, nothing more.

Washington trained brickmakers, carpenters, and machinists there instead. Ordinary people the world had written off became master tradespeople, and their work became renowned across the country. His own words still hold:

“Excellence is to do a common thing in an uncommon way.”

That’s the real opportunity hiding inside the AI skills gap. You don’t need rare people to get rare results. You need the people already on your payroll doing their common, everyday tasks with real, uncommon skill.

The Skills That Actually Transfer

The OECD frames workforce AI skills as three groups:

• Foundational skills everyone needs to function in a digital economy
• ICT skills for the people who build and maintain AI systems
• Complementary skills, the judgment, communication, and process thinking that make AI genuinely useful inside a real job.

That third group carries more weight than most hiring plans assume. Across the occupations most exposed to AI, 72% of vacancies call for management skill and 67% call for business-process skill, both well ahead of demand for technical AI skill itself.

Put plainly, employers are already competing for skills a well-run team has spent years building: communication, judgment, and process ownership, just now applied to AI-shaped work. Your 99% already carry a good share of what the market is asking for.

Ending Thought:

The AI specialist myth is comforting because it makes the AI skills gap sound like a recruiting problem. In practice, it’s a capability-building problem, and one most organizations already have the people to solve.

The 99% of your workforce who will never be AI specialists are exactly the people who can turn AI literacy into real, everyday advantage, if you give them a clear path to build it. If you’re not sure whether your organization has a hiring gap or a training gap, Radiant Institute can help you find out. Reach out to explore how our AI enablement training programs can close the gap with the people you already have.