73% of Small Businesses Want More AI Training — The Competitive Gap

A new survey from Goldman Sachs landed in June 2026, and the headline number should shake every small business owner in the United States. The 2026 Small Business AI Survey, conducted across more than 1,400 US business owners and executives, found that 76% of American small businesses have adopted AI tools in some capacity. Among businesses with 10-50 employees, adoption climbs to 82%.

But adoption and proficiency are not the same thing. The number that matters: 73% of respondents say they need “substantially more” training to use AI effectively.

Seventy-three percent of business owners who already bought AI tools admit they do not know how to use them properly. That is not a technology problem. It is a training problem. And it is creating a growing divide between businesses that treat AI as a skill to learn and businesses that treat AI as an app to install.

What the Goldman Sachs Survey Actually Measured

The survey divided US small business AI engagement into three tiers. The breakdown reveals where the market really stands in mid-2026.

• Tier 1 — Basic tool users (38% of respondents): These businesses use ChatGPT, Claude, or Copilot for simple tasks — drafting emails, summarizing documents, basic research. They have adopted AI but have no structured training plan and limited understanding of prompt engineering, retrieval-augmented generation, or model selection.
• Tier 2 — Intermediate adopters (34%): These businesses use multiple AI tools across departments — marketing, accounting, customer service. They have experimented with custom instructions and basic workflow automation. Most are self-taught through YouTube tutorials and trial-and-error.
• Tier 3 — Advanced deployers (28%): These businesses treat AI as an infrastructure investment. They have designated AI leads, structured training budgets, documented workflows, and measurable ROI tracking. This group reports 43% higher revenue growth than Tier 1 businesses over the past 12 months.

The training gap is concentrated in Tiers 1 and 2, which together represent 72% of US small businesses. These are not businesses that reject AI. They are businesses that bought AI tools and then realized the tools are only as good as the people operating them.

The High Cost of Untrained AI Use

Goldman Sachs found that businesses that invested in structured training alongside tool adoption report 40% higher satisfaction with their AI spending. Those that bought tools without any formal training are 2.3 times more likely to say AI has “not delivered measurable results.”

Untrained AI use creates specific, measurable problems:

• Hallucinated outputs accepted as fact: Employees who do not understand model limitations treat AI-generated information as authoritative. A Goldman Sachs case study cited a US retailer that lost $12,000 on a single inventory decision because an untrained employee trusted a hallucinated product demand projection.
• Wasted subscription spend: The survey found that 41% of small businesses pay for three or more AI tools but actively use two or fewer. Training reduces tool sprawl because trained users consolidate workflows into fewer, better-integrated tools.
• Security exposure: Untrained employees are 3x more likely to paste sensitive data into public AI interfaces. The U.S. Chamber of Commerce, which partnered with Goldman Sachs on the survey, identified data leakage as the fastest-growing small business AI risk in 2026.

In the US market, these problems compound. A business with five employees each making $15/hour wastes roughly $15,600 annually in lost productivity from untrained AI use, according to the survey’s estimates. That is the equivalent of one full-time employee’s salary at current minimum wages in most states.

What 73% of Small Businesses Actually Need

The training demand breaks down into four areas, ranked by priority across survey respondents:

1. Prompt engineering and tool-specific skills (cited by 68%): Most small business owners who use ChatGPT, Claude, or Gemini have never learned how to write an effective prompt. The difference between “write an email” and “write a follow-up email to a client who ghosted after the third meeting, referencing our June pricing discussion, with a tone that rebuilds confidence without sounding desperate” is the difference between a generic output and a usable one. Goldman Sachs found that businesses investing in at least 8 hours of prompt engineering training per employee saw 3x higher satisfaction with AI outcomes.
2. Workflow integration (cited by 54%): Business owners who use AI in isolation — generating text in ChatGPT, then manually pasting it into a CRM, then manually triggering an email — capture only a fraction of AI’s value. Training on API connectivity, Zapier-style automation, and embedded AI features within existing tools like Salesforce, HubSpot, and QuickBooks was the second-highest priority.
3. Output verification and fact-checking (cited by 47%): As AI models improve, outputs become more convincing — but not more accurate. Training on how to verify AI-generated facts, cross-reference sources, and identify hallucination patterns was cited as a critical need. The U.S. Chamber of Commerce’s free AI training program includes a module on verification that Goldman Sachs specifically recommends.
4. Strategic application (cited by 39%): The highest tier of training demand came from businesses that already know how to use AI tools but do not know where to apply them for maximum business impact. These owners wanted case studies, industry-specific playbooks, and guidance on building AI into their actual business model rather than just using it as a productivity aid.

The Two-Tier Market Is Already Here

Goldman Sachs stratified its results by training investment and found a clear divergence. Businesses that invested 10-15% of their total AI budget in training reported 43% higher revenue growth from AI initiatives than those that invested less than 5%. The gap is not small. It is the difference between AI being a cost center and AI being a profit center.

Among the advanced deployers in Tier 3, the average training investment was $1,200 per employee per year. Among Tier 1 businesses, the average was $87. The difference in outcomes correlates directly: Tier 3 businesses reported an average of $23,000 in annual AI-driven cost savings per employee, while Tier 1 businesses reported $2,100. The 14x gap in training spend produced an 11x gap in returns.

In a competitive market with thin margins, that gap compounds over time. A business that invests $1,200 per employee this year captures $23,000 in savings — a 19x return. The business that spends $87 captures $2,100 — a 24x return on paper, but on an absolute basis it is $21,000 less. Over three years, the gap exceeds $60,000 per employee. For a 10-person business, that is $600,000 in total value left on the table.

Why Structured Training Beats Self-Taught AI

The temptation for most small business owners is to treat AI like a smartphone — buy it, open it, figure it out as you go. That approach works for Instagram. It does not work for generative AI, where the difference between a good and bad output depends entirely on the quality of the input and the sophistication of the evaluation criteria applied to the output.

Goldman Sachs found that self-taught AI users spend an average of 3.2 hours per week “fixing” AI-generated outputs — correcting errors, rewriting text, verifying facts. Structured training reduces that to 1.1 hours per week. Over a year, that is 109 hours of reclaimed time per employee — nearly three full work weeks.

The ROI calculation is straightforward for US small business owners. An employee earning $25/hour costs the business roughly $52,000 annually including overhead. If structured training reclaims 109 hours per year, that is $2,725 in recovered productivity per employee. For a 10-person business, training five core employees costs roughly $5,000 and recovers $13,625 in productivity in year one alone. In year two, the recovery is full without additional training cost.

This math works because training is a one-time investment with recurring returns. AI subscriptions recur forever. The businesses that understand this asymmetry are the ones pulling ahead.

Industry-Specific AI Training: Where US Businesses Need Help Most

The Goldman Sachs survey broke down training demand by industry vertical, revealing significant variation in what American small businesses actually need:

• Professional services (legal, accounting, consulting): 81% want training on AI accuracy verification and hallucination detection. The cost of an AI error in contract analysis or tax preparation is substantially higher than in marketing copy, so these businesses demand verifiable outputs.
• Retail and e-commerce: 76% want training on AI-driven inventory management and customer personalization. These businesses already use AI for product descriptions and ad copy but struggle to connect AI outputs to actual sales data.
• Healthcare and wellness: 72% want training on HIPAA-compliant AI use and patient data handling. The regulatory layer adds complexity that off-the-shelf AI tools do not address.
• Construction and trades: 68% want training on AI-powered estimating and project management. These are among the fastest-growing AI adopters in 2026, according to Goldman Sachs, starting from a very low training base.

The takeaway: one-size-fits-all AI training does not work. A plumber and a CPA need fundamentally different AI skills. The businesses that succeed are the ones seeking industry-specific training rather than generic “AI for business” courses.

nn

Free Training Options That Actually Work

The U.S. Chamber of Commerce, the Small Business & Entrepreneurship Council, and most major AI providers offer free training resources. The U.S. Chamber of Commerce AI training program includes structured modules on prompt engineering, AI ethics, and workflow integration — all free and targeted specifically at small business owners. OpenAI, Anthropic, and Google each offer certification programs for their respective tools.

The key, according to Goldman Sachs, is structured learning paths rather than random YouTube tutorials. Provider-specific certifications aligned to the tools actually in use outperform general “AI literacy” courses by a 2:1 margin in measured business outcomes.

Frequently Asked Questions About AI Training for Small Businesses

How much should a small business spend on AI training?

Goldman Sachs recommends allocating 10-15% of your total AI budget to training and upskilling. For a business spending $500/month on AI tools, that translates to $50-75/month in training — often available through free resources like the U.S. Chamber of Commerce program.

What are the top AI training topics for small businesses?

Prompt engineering (cited by 68%), workflow integration (54%), output verification and fact-checking (47%), and strategic application (39%) are the four most requested training areas according to the Goldman Sachs 2026 Small Business AI Survey.

How long does it take to train employees on AI?

Basic AI literacy takes 4-6 hours of structured training. Proficiency with specific tools takes 20-40 hours over 4-6 weeks. Businesses investing in at least 8 hours of training per employee saw 3x higher satisfaction with AI outcomes.

Are there free AI training resources for small businesses?

The U.S. Chamber of Commerce, SBE Council, and most major AI tool providers (OpenAI, Anthropic, Google) offer free training resources. Structured learning paths outperform ad-hoc YouTube tutorials by a 2:1 margin.