AI Was Supposed to Mean Less Work. Two New Surveys Say Small Businesses Are Hiring and Working More.

Two surveys published this month say much the same thing from different angles. On 2 October the U.S. Chamber of Commerce released its fifth report on technology and small businesses. On 14 October the headset maker Jabra published a global study of more than 6,000 office workers. Neither found AI making work disappear. Both found it making work change, and often grow.

The Chamber: more small businesses use AI, and most say it is adding jobs

• 66% of U.S. small businesses now use AI, up from 23% in 2023 and up 8 points on last year.
• 47% say AI is creating jobs in their business today. 6% say it is letting them cut staff.
• Small businesses that use AI are at least 7 points more likely to report growth in sales, profit and headcount than those that do not.
• 95% of AI-using businesses say they are training staff to use it.
• 78% say losing access to AI would slow their growth.
• Six in ten worry about the cost of complying with different AI and privacy laws in different states.

Jabra: workers say the work grew, especially for heavy AI users

Jabra's study was run by the research firm Toluna across the US, the UK, Germany, France, Poland and India. It asked people how their time had changed over the past year:

• 85% said at least one kind of work had increased. Only 35% said any kind had decreased.
• Among people who use AI several times a day, 57% said their admin and coordination work went up, against 38% of everyone else.
• Heavy users were also more likely to say strategic planning went up (56% against 35%) and creative work went up (56% against 30%).
• 64% of heavy users said they collaborate with colleagues more than a year ago, against 25% of light users.

Read them with care

Both are surveys of what people say, not measurements of hours or payroll. Both come from organisations with an interest in the answer. The Chamber is a business lobby that is campaigning against a patchwork of state AI rules, and "AI creates jobs" supports that case. Jabra sells headsets and conferencing equipment, and its report also promotes voice AI and meeting hardware. Jabra's own figures show that heavy AI users report more work. They do not show that AI caused it. Busier people may simply be the ones who reach for AI.

Still, the two point the same way, and they match what we see when we put AI into a real workflow.

Where the time goes

AI is very good at the first draft: the email, the quote, the summary, the code, the report. That step really does get faster. But a first draft is rarely the end of the job. Someone still has to check it, fix what is wrong, send it to the right person, chase the answer and decide what happens next.

When drafting gets ten times faster, there are ten times as many drafts to check. That is the admin and coordination growth in Jabra's numbers. When a small team can suddenly take on more work, it often needs another person to handle it. That is the hiring in the Chamber's numbers. Neither is a failure of AI. Both are what happens when the slow step in a process moves.

Put simply: AI made writing the first version cheap. It made checking it the expensive part.

How to plan an AI rollout around the work that grows

1. Measure the whole job, not the AI step. Time a quote, an invoice or a support reply from the request coming in to the customer getting the answer. If only the drafting time drops, the job has not got much faster.
2. Budget time for checking. Decide up front who reviews AI output, how long it should take, and what they are checking for. Write it down. Unplanned checking is where the hours pile up.
3. Make the output easy to check. Ask for sources, totals that add up, and a list of anything the AI was unsure about. A draft that shows its working takes minutes to review. One that does not can take longer than writing it yourself.
4. Remove a step, do not just speed one up. The biggest gains come when AI lets you drop a handover, a re-typing step or an approval that only existed because the old process was slow.
5. Give one person ownership of each AI workflow. Someone who notices when the output gets worse, the cost goes up or the tool changes. Without an owner, problems turn into everyone's extra work.

We wrote about the same pattern from the other side in the automation trap, and about which accounting tasks are safe to hand over in what accounting firms can give to AI.

The short version

The surveys do not say AI is a disappointment. Small businesses using it report more growth, and most of them would not give it up. They say the time saved does not come back as free time. It comes back as more work of a different kind. Plan for that work, and AI pays off. Plan for fewer hours, and the checking lands on whoever is busiest.

Sources

• U.S. Chamber of Commerce Technology Engagement Center, Empowering Small Business: The Impact of Technology on U.S. Small Business, 2 October 2026 (fifth edition). AI adoption 66% (23% in 2023), 47% job creation vs 6% reductions, the 7-point growth gap, 95% training, 78% on losing access, six in ten worried about state laws.
• IT Brief, Jabra study finds AI adds work & boosts collaboration, 14 October 2026. Jabra's study by Toluna: 6,000+ knowledge workers in six countries, 85% vs 35%, heavy-user figures for admin, strategic, creative work and collaboration.
• KSWO, AI creating more jobs than it's eliminating at small businesses, US Chamber research finds, 23 October 2026. The Chamber's Jordan Crenshaw on the 47% and 6% figures.