Everyone Is Using AI? Census Data Says About 1 in 5 Businesses
If a podcast told you AI is about to change everything, the next call you get will probably be someone selling it to you. Before you take that call, here's what the numbers actually say.
Tech podcasts are full of confident predictions about AI. Some of it is true. Much of it is a guess said with a straight face. Those guesses don't stay on the podcast. They show up in your inbox as a sales pitch.
Picture an HVAC owner in Costa Mesa or a roofer in Anaheim. They hear on a Tuesday that AI will run their business. By Friday, someone is pitching them a $5,000 "AI marketing system." Here are the five claims that make that sale work, run through our desk.
CLAIM: "Everyone is already using AI. You're falling behind."
VERDICT: FALSE (for small businesses)
FINDING: The U.S. Census Bureau's Business Trends and Outlook Survey is the biggest government measure of business AI use. From December 2025 to May 2026, overall AI use held between 17% and 20% of businesses. The gap by size is sharp. Among firms with 250 or more employees, 37% used AI. Among firms with four or fewer employees, less than 20% did. And for firms under 20 employees, use didn't change in any meaningful way over those six months.
Source check: The survey is funded by the federal government, not a vendor. Its sample is about 1.2 million businesses, surveyed every two weeks. The data is from 2026, so it's current. One caution: in November 2025 the Census Bureau broadened the question to cover AI use in any part of the business, not just in making goods or services. So the numbers before and after that change aren't directly comparable.
Evidence tier: Sourced/Verified.
RECOMMENDATION: When a vendor says "everyone's doing it," ask them who "everyone" is. For a crew your size, the honest answer is "most of your competitors aren't yet." Large Firms With at Least 20 Employees Biggest AI Users +3
The people telling you everyone's already doing it are usually the people selling it.
CLAIM: "Studies show most small businesses use AI now."
VERDICT: HYPE
FINDING: How you ask the question changes the answer. One analysis notes that the U.S. Chamber of Commerce reported 58% of small businesses use generative AI, compared with the Census figure of about 20%. A survey that counts asking ChatGPT to write one email gets a big number. A survey that asks whether AI is part of how you run the business gets a small one. Vendors quote whichever number sells.
Evidence tier: Directional. The 58% figure is secondhand, and we did not review the Chamber's methods directly.
RECOMMENDATION: Any time you hear a percentage, ask: "Using AI to do what?" Omago
CLAIM: "AI will create way more jobs than it takes away."
VERDICT: HYPE
FINDING: This is a prediction dressed up as a fact. The best-known forecast comes from the World Economic Forum. It projects 170 million new jobs and 92 million lost by 2030, for a net gain of 78 million. But that forecast covers technology, the economy, population shifts, and the green transition together, not AI alone. It's also a survey of what employers expect, not something that has happened. Even the speaker in one widely shared AI video later admitted nobody knows whether AI will destroy jobs.
Evidence tier: Directional.
RECOMMENDATION: Treat big job predictions as opinions. They don't tell you anything about your next hire or your next invoice. sabanciuniv
CLAIM: "If a famous tech investor says it, it's been checked."
VERDICT: FALSE
FINDING: In that same widely shared video, the speaker said the Industrial Revolution began in 1910. It began in Britain in the 1700s. He said Nelson Rockefeller paid for the schools. It was John D. Rockefeller Sr. He started the General Education Board in 1902 with $1 million, and nearly $325 million of his money eventually went through it. He did get Andrew Carnegie right. Carnegie and his foundation spent about $56 million to build 2,509 public libraries, 1,681 of them in the United States. That's one fact right and two wrong, said with the same confidence. Nobody in the room stopped him.
Evidence tier: Sourced/Verified.
RECOMMENDATION: Fame isn't a source. Check the claim, not the person saying it. philanthropyroundtablecarnegie
CLAIM: "Our AI system runs your marketing for you."
VERDICT: HYPE
FINDING: Even the AI enthusiasts don't believe this. In the same video, the speaker argues the part people can't hand off is "taste and judgment": deciding what's true, what fits, and what to do next. His example is a loan officer. AI processes the loan, but a human still decides yes or no. A system that promises to take the judgment off your plate is promising something its own supporters say it can't do.
Evidence tier: Experience-Based. This is the speaker's own argument, and no data measures it.
RECOMMENDATION: Ask the vendor which decisions stay with you. If the answer is "none," walk.
WHAT TO ASK BEFORE YOU BUY ANY "AI" MARKETING SERVICE
Show me a business my size, in my trade, that got results from this, and let me call them.
What does the AI actually do, and what does a person do?
If this doesn't work in 60 days, what do I owe you?
A real vendor answers all three without flinching. Anyone who changes the subject has told you what you need to know.
Want a second set of eyes on an AI pitch that landed on your desk? Book 15 minutes: https://calendly.com/kirby-thelocalaim/15min
AI is a tool, not a verdict. Make whoever's selling it prove it works for a business your size, in writing, with numbers.