The AI Skills Gap in Small Business: What the Data Shows and What to Do

The AI skills gap in small business is the distance between how many SMBs have adopted an AI tool and how many have the policies, training, and role redesign needed to use it safely and well — and recent survey data shows that gap is wide. A TriNet and Harvard Business Review Analytic Services survey of 230 U.S. SMB respondents, conducted in early 2026, found that just 19% felt their organization was highly prepared to recruit or develop the AI skills it needs, according to the survey coverage reported by Barchart. Adoption has outrun readiness, and that mismatch is now the more pressing problem.

What the Gap Actually Looks Like Right Now

The pattern shows up consistently across multiple surveys, not just one. A 2026 survey of more than 300 member businesses by the Employers and Manufacturers Association found 83% of respondents were already using AI in some form, yet only 13% had a formal AI policy in place, and just 4% said AI initiatives were being led by HR — the function most directly responsible for workforce capability (EMA survey findings). Put plainly: the tools are in, but almost nobody has written down who's allowed to use them for what, and the department built to manage workforce capability usually isn't in the room.

This isn't a story about SMBs being behind on buying AI tools — adoption numbers are consistently high across these surveys. It's that adoption and readiness have become two separate curves, and most small businesses have only moved on the first one.

Why This Is a Governance Problem, Not Just a Training Problem

It's tempting to read "skills gap" as "our people need a class." Training is part of the answer, but the EMA data points at something structural: an 83%-adoption, 13%-policy ratio means most AI use inside these businesses is happening without anyone having decided what's in bounds. That's not a knowledge gap in the workforce — it's a decision that leadership hasn't made yet. No policy means no answer to basic questions: can a staff member paste customer data into a public chatbot? Who approves a new AI tool before it touches a live workflow? What happens when the tool gets something wrong in front of a customer?

Recognizing AI use that's already happening without sanction or structure — sometimes called shadow AI — is a related risk we've covered in more depth in our piece on shadow AI risk for small business. The skills gap and the shadow-AI problem are two faces of the same root cause: tools adopted faster than governance caught up.

The Three Places the Gap Shows Up First

1. Hiring and role design. Job descriptions and interview processes mostly haven't changed to reflect that a role now includes working alongside AI tools day to day. That makes it hard to hire for the skill and hard to evaluate whether an existing employee has it.
2. Workflow redesign. Buying a tool isn't the same as redesigning the workflow around it. Most of the value AI tools promise only shows up when someone restructures how a task gets done, not when the tool is simply bolted onto the old process.
3. Policy and oversight. As the EMA numbers show, this is the most commonly skipped step, and it's the cheapest one to fix relative to its risk reduction.

Closing the Gap Without a Big Training Budget

The temptation is to treat this as a large initiative requiring a training vendor and a multi-month rollout. For most SMBs, the higher-leverage starting point is smaller and more structural:

• Write the policy first, even a short one. A one-page AI acceptable use policy — what tools are approved, what data can and can't go into them, who approves a new one — closes the biggest gap the EMA survey identifies, and it costs almost nothing to produce. We've laid out a practical starting template in our post on AI acceptable use policy for small business.
• Train inside real workflows, not in the abstract. Generic AI literacy sessions build less actual capability than role-specific, hands-on training where staff use the tool on their own real tasks. Our piece on AI training for employees goes deeper on what that looks like in practice — this post is about the broader organizational gap; that one is about closing it person by person.
• Assess before you expand. Running a structured readiness check before adding the next AI tool catches the gap while it's still small and specific, rather than after three more tools have been adopted without a plan.

What Government and Industry Are Watching

The gap is visible enough now that it's drawing policy attention. A bill reintroduced in the U.S. Senate in February 2026 by Senators Cantwell and Moran would direct the Department of Commerce and the Small Business Administration to build AI training resources specifically for small firms, with sponsors citing a lack of technical expertise and structured training as the core obstacle, according to the Senate Commerce Committee's announcement. Whether or not that specific bill becomes law, it's a signal that the readiness gap identified in these surveys is being treated as a real economic concern, not a niche HR issue.

What This Means for Hiring and Performance Reviews

One consequence of this gap that gets less attention than policy and training: performance expectations and hiring criteria haven't caught up either. If a role now routinely involves working alongside an AI tool, that should show up in how the role is evaluated and how candidates are screened — but for most SMBs, job descriptions and review criteria were written before AI tools were part of daily work, and nobody has gone back to update them. That leaves managers evaluating performance against an outdated picture of what the job actually involves, and it leaves new hires guessing at an expectation nobody wrote down.

This doesn't require an overhaul. It's closer to a maintenance task: the next time a job description or review template comes up for its normal update cycle, add a line about the AI tools relevant to that role and what competent use looks like. Treating it as a one-time, dedicated HR project is how it ends up permanently deferred — the same failure pattern the EMA data shows for policy work generally.

Common questions

Is the AI skills gap mainly about employees not knowing how to use AI tools? Partly, but the bigger structural gap is organizational: most SMBs haven't written an AI policy, assigned ownership of AI governance, or redesigned roles around AI use. Individual tool literacy is real but it's the smaller half of the problem the survey data points to.

Do we need to hire an AI specialist to close this gap? Not usually, at least not first. The EMA and TriNet/HBR data both suggest the missing piece is policy and structured planning, which existing leadership can put in place, rather than a new technical hire. A specialist becomes worth considering once AI use is central enough to the business to need dedicated ongoing oversight.

What should we do first if we have no AI policy at all? Write a short one before you add anything else. Define which tools are approved, what data is off-limits for any AI tool, and who signs off on a new one. That single document addresses the gap the EMA survey found in 87% of surveyed businesses, and it takes days, not months.

Is this gap specific to any one industry? The surveys cited here span general SMB populations across the US, UK, and New Zealand, so the pattern isn't industry-specific — it shows up anywhere adoption has moved faster than governance, which describes most small businesses that have added an AI tool in the last two years.

Closing this gap starts with an honest look at where your business actually stands — not where the tool vendor's demo suggested you'd be. Start a systems audit and we'll help you find the policy, training, and workflow gaps before they turn into a bigger problem.